Built-In Zoo Models#
This page lists all of the natively available models in the FiftyOne Model Zoo.
Check out the API reference for complete instructions for using the Model Zoo.
Classic neural network that recognizes images and helped launch the deep learning revolution
CenterNet model from "Objects as Points" with the Hourglass-104 backbone trained on COCO resized to 1024x1024
CenterNet model from "Objects as Points" with the Hourglass-104 backbone trained on COCO resized to 512x512
CenterNet model from "Objects as Points" with the MobileNetV2 backbone trained on COCO resized to 512x512
CenterNet model from "Objects as Points" with the ResNet-101v1 backbone + FPN trained on COCO resized to 512x512
CenterNet model from "Objects as Points" with the ResNet-50-v1 backbone + FPN trained on COCO resized to 512x512
CenterNet model from "Objects as Points" with the ResNet-50v2 backbone trained on COCO resized to 512x512
Vision transformer for image classification and custom fine-tuning on specialized datasets
Understands both images and text together, enabling search and classification using natural language descriptions
Base modern CNN with transformer elements for robust visual understanding
Large modern CNN demonstrating competitive performance with vision transformers
Small modernized CNN delivering strong accuracy through architectural innovations
Tiny modern CNN bridging traditional convolutions with transformer-inspired improvements
Extra-large modern CNN maximizing architectural improvements for top accuracy
DeepLabv3+ semantic segmentation model from "Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation" with Xception backbone trained on the Cityscapes dataset
DeepLabv3+ semantic segmentation model from "Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation" with MobileNetV2 backbone trained on the Cityscapes dataset
Labels everyday objects in images pixel by pixel for general scene understanding and analysis
Faster version that quickly identifies and labels objects in images for real-time applications
Compact yet powerful classifier that delivers strong results while using minimal computational resources
Dense network that achieves high accuracy for image classification and adapts well to new tasks
Deeper variant offering improved accuracy while remaining efficient enough for practical deployment
Extra-deep model providing the most detailed features for complex image understanding tasks
Hugging Face Transformers model for monocular depth estimation
Modern object detector that finds items in images without needing complex post-processing steps
D-FINE Large from "D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement" trained on COCO. Achieves 54.0% AP at 124 FPS on T4 GPU.
D-FINE Medium from "D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement" trained on COCO. Mid-size real-time object detector.
D-FINE Nano from "D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement" trained on COCO. Ultra-lightweight real-time object detector.
D-FINE Small from "D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement" trained on COCO. Balanced real-time object detector.
D-FINE XLarge from "D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement" trained on COCO. Achieves 55.8% AP at 78 FPS on T4 GPU.
Enhanced image search model that resists noise and errors for more reliable similarity matching
Creates searchable image fingerprints for finding similar pictures and organizing large photo collections
Highest-capacity dinov2 search model with maximum stability for finding images across massive diverse datasets
Powerful image search engine that handles enormous photo collections with rich detail extraction
Large stable model for finding and grouping similar images across big databases reliably
Large model that creates detailed image fingerprints for advanced search and automatic grouping
Compact stable model for image search that runs efficiently on phones and edge devices
Small model enabling image search and similarity matching directly on mobile devices
EfficientDet-D0 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 512x512
EfficientDet-D0 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO
EfficientDet-D1 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 640x640
EfficientDet-D1 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO
EfficientDet-D2 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 768x768
EfficientDet-D2 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO
EfficientDet-D3 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 896x896
EfficientDet-D3 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO
EfficientDet-D4 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 1024x1024
EfficientDet-D4 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO
EfficientDet-D5 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 1280x1280
EfficientDet-D5 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO
EfficientDet-D6 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 1280x1280
EfficientDet-D6 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO
EfficientDet-D7 model from "EfficientDet: Scalable and Efficient Object Detection" trained on COCO resized to 1536x1536
Efficient image classifier optimized for mobile devices with excellent accuracy-efficiency tradeoff
Scaled efficient classifier with improved accuracy for slightly larger computational budgets
Balanced efficient model providing stronger performance while maintaining reasonable resource usage
Mid-scale efficient classifier delivering high accuracy for versatile deployment scenarios
Large efficient model with enhanced features for transfer learning applications
High-capacity efficient classifier prioritizing accuracy with available compute resources
Extended efficient model approaching state-of-the-art accuracy on challenging datasets
Maximum efficient classifier pushing performance boundaries while preserving efficiency principles
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" atrous version with Inception backbone trained on COCO
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" atrous version with low-proposals and Inception backbone trained on COCO
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" with Inception v2 backbone trained on COCO
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" with NAS-net backbone trained on COCO
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" with low-proposals and NAS-net backbone trained on COCO
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" with ResNet-101 backbone trained on COCO
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" with low-proposals and ResNet-101 backbone trained on COCO
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" with ResNet-50 backbone trained on COCO
Multi-scale object finder that accurately detects both small and large items in images
Faster R-CNN model from "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks" with low-proposals and ResNet-50 backbone trained on COCO
Creates detailed pixel-level labels for images, identifying and outlining twenty-one different object categories
Fast image labeler that quickly identifies and outlines objects for interactive editing and annotation
Classic image classifier providing reliable categorization and features for various computer vision projects.
Hugging Face Transformers model for zero-shot semantic segmentation
Inception v2 model from "Rethinking the Inception Architecture for Computer Vision" trained on ImageNet
Efficient image classifier delivering accurate results with useful features for transfer learning applications
Inception v4 model from "Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning" trained on ImageNet
Finds people in images and maps their body joints for pose estimation and motion analysis
Mask R-CNN model from "Mask R-CNN" atrous version with Inception backbone trained on COCO
Mask R-CNN model from "Mask R-CNN" with Inception backbone trained on COCO
Mask R-CNN model from "Mask R-CNN" atrous version with ResNet-101 backbone trained on COCO
Mask R-CNN model from "Mask R-CNN" atrous version with ResNet-50 backbone trained on COCO
Multi-scale object outliner using advanced architecture for accurate segmentation across different object sizes
Medical segmentation tool that outlines organs and structures in medical videos and 3D scans
Ultra-lightweight image classifier designed by AI for running directly on phones and IoT devices
Mobile-optimized classifier balancing size and accuracy for efficient on-device image recognition
MobileNetV2 model from "MobileNetV2: Inverted Residuals and Linear Bottlenecks" trained on ImageNet
Mobile-friendly image classifier optimized for quick training and deployment on resource-limited devices
Real-time detector that finds any object you describe in words, perfect for live video analysis
Connects images with text descriptions enabling search by words and automatic content filtering systems
MiniCPM-V 4.5 is the latest and most capable model in the MiniCPM-V series. The model is built on Qwen3-8B and SigLIP2-400M with a total of 8B parameters.
Finds any object you name in pictures using 16x16 image patches without needing specific training for those items
Finds any object you name in pictures using efficient 32x32 image patches without needing specific training
Large OWL-ViT zero-shot object detector with ViT-L/14 backbone. Achieves higher accuracy than base models, especially for smaller objects.
ResNet-50 v1 model from "Deep Residual Learning for Image Recognition" trained on ImageNet
ResNet-50 v2 model from "Deep Residual Learning for Image Recognition" trained on ImageNet
Deep image recognition model delivering high accuracy for demanding classification and analysis tasks
Very deep classifier providing the richest visual features for precision-critical image understanding applications
Lightweight image classifier designed for fast recognition on phones and other resource-limited devices
Balanced image classifier offering good accuracy and speed for everyday computer vision needs
Most popular image recognition backbone widely used as starting point for custom vision projects
Powerful image classifier with enhanced capacity for handling complex visual recognition challenges effectively
Efficient advanced classifier delivering strong accuracy with reasonable computing requirements for practical deployments
Fast object detector that quickly finds and boxes eighty common items in any image
R-FCN object detection model from "R-FCN: Object Detection via Region-based Fully Convolutional Networks" with ResNet-101 backbone trained on COCO
Modern real-time object detector that finds items without complex post-processing for responsive applications
Balanced real-time object detector offering improved accuracy for production use
Lightweight real-time object detector optimized for speed on edge devices
High-capacity object detector delivering very precise results at speeds suitable for production use
Efficient transformer-based semantic segmentation model for scene parsing with 150 classes
Balanced SegFormer model providing good accuracy-efficiency tradeoff for scene understanding
Medium-sized SegFormer delivering enhanced segmentation quality for complex scenes
Larger SegFormer model with improved accuracy for detailed semantic segmentation
High-capacity SegFormer achieving excellent results on challenging segmentation tasks
Largest SegFormer model delivering the best semantic segmentation performance in its family
Accurate image segmentation model for editing, labeling, and creative work with still pictures
Video segmentation model that tracks and outlines objects throughout clips for editing and analysis
High-quality image segmenter producing detailed masks for demanding professional editing and annotation tasks
Advanced video segmenter providing fine object tracking throughout full videos for post-production work
Fast image segmentation model that runs efficiently on laptops and edge computing devices
Quick video segmentation model delivering rapid object tracking on standard graphics cards
Smallest image segmentation model offering instant results for mobile apps and embedded systems
Tiny video segmenter enabling real-time object tracking on phones and compact devices
Updated image segmenter with improved mask accuracy for everyday editing and dataset creation
Enhanced video segmenter with better tracking quality for video analysis and scene understanding
Large updated model offering even finer masks for high-resolution professional image workflows
Large video model producing exceptionally detailed masks throughout long videos for intensive production
Balanced updated segmenter combining speed and accuracy for edge device image processing
Improved video segmenter maintaining quick performance on compact hardware while enhancing mask quality
Enhanced mobile image segmenter for apps, augmented reality filters, and on-device processing
Upgraded mobile video segmenter for live effects on phones, wearables, and smart cameras
Interactive segmentation tool that instantly outlines any object you point to or describe
Highest quality segmentation model creating extremely detailed masks for research and large-scale annotation projects
Large segmentation model producing finer object outlines for professional editing and labeling workflows
Hugging Face Transformers model for semantic segmentation
Ultra-small image classifier for tiny devices with very limited power and memory
Mobile image classifier that works efficiently on phones with modest computing resources
Hugging Face Transformers model for zero-shot image classification
Tiny image classifier that fits in just five megabytes for embedded devices
Ultra-compact image classifier perfect for severely resource-constrained hardware and applications
Inception Single Shot Detector model from "SSD: Single Shot MultiBox Detector" trained on COCO
Single Shot Detector model from "SSD: Single Shot MultiBox Detector" with MobileNetV1 backbone trained on COCO
MobileNetV1 model from "MobileNetV2: Inverted Residuals and Linear Bottlenecks" resized to 640x640
FPN Single Shot Detector model from "SSD: Single Shot MultiBox Detector" with MobileNetV1 backbone trained on COCO
MobileNetV2 model from "MobileNetV2: Inverted Residuals and Linear Bottlenecks" resized to 320x320
FPN Single Shot Detector model from "SSD: Single Shot MultiBox Detector" with ResNet-50 backbone trained on COCO
Base hierarchical transformer delivering strong results across vision tasks
Large hierarchical transformer with enhanced capacity for demanding applications
Small hierarchical transformer balancing efficiency and performance for practical use
Tiny hierarchical transformer for efficient visual recognition on edge devices
Classic image classifier with stable training useful for various computer vision projects
Simple baseline image classifier valuable for research experimentation and learning purposes
Deeper classic classifier providing stable training process and solid accuracy results overall
Straightforward image classifier valued for easy experimentation and model compression studies
Popular feature extractor widely used for detection, style transfer, and medical imaging
VGG-16 model from "Very Deep Convolutional Networks for Large-Scale Image Recognition" trained on ImageNet
PyTorch version of the popular classifier ready for modern deep learning workflows
Deep classic model providing rich features for style transfer and interpretability analysis
Deep image classifier delivering detailed features for creative applications and research projects
Modern image classifier that recognizes objects and provides useful features for various computer vision tasks
Extra-wide deep classifier for high-precision image recognition and advanced transfer learning tasks
Wide classifier offering stronger accuracy and better features for adapting to new tasks
AI-designed detector family offering three model variants for diverse deployment scenarios
YOLOv2 model from "YOLO9000: Better, Faster, Stronger" trained on COCO
Real-time object detector balancing high accuracy with fast processing speeds effectively
Model creating detailed object outlines for precise image editing and analysis
Object detector offering good balance between speed and accuracy for most applications
Model generating object masks efficiently for everyday segmentation tasks
Object detector designed specifically for phones and other edge computing devices
Edge model producing object outlines directly on phones and edge devices
Fast object detector ideal for systems with limited graphics processing power
Model creating object masks quickly for real-time segmentation applications
Object detector prioritizing accuracy over processing speed for critical applications
Model delivering high-quality object outlines for professional workflows
Real-time model creating both object outlines and boxes for any described item
Model producing masks and boxes for objects described in natural language
Segments specified classes, generating object outlines and boxes for real-time applications
Model outlining and boxing any object you describe without specific training
Model creating masks for any object type you name in text
Compact model producing outlines for objects described in words on edge devices
Object detector with special optimizations for even faster inference on modern hardware
Balanced detector providing good accuracy and speed for general-purpose object detection tasks
Edge-optimized detector for devices with minimal computing resources available
Fast lightweight detector suitable for systems with limited GPU capabilities and memory
High-accuracy detector for demanding object detection applications and research
Real-time detector producing accurate results quickly for demanding vision applications
Real-time detector balancing good accuracy with fast processing speeds
Lightweight detector for edge devices needing basic object detection capabilities
Real-time detector delivering good results with minimal computational requirements
High-accuracy detector offering top precision for applications where quality is critical
Real-time detector with advanced architecture for improved object finding in complex scenes
Specialized detector that finds rotated objects in aerial and satellite imagery accurately
General-purpose detector trained on diverse images recognizing over six hundred object categories
Creates precise object outlines for detailed image editing and analysis tasks
Finds and boxes any object you describe using natural language prompts
Detector balancing speed and accuracy for everyday object detection needs
Finds rotated bounding boxes in aerial images for mapping and surveillance applications
Versatile detector recognizing hundreds of different object types across varied image domains
Generates object masks with good balance of speed and quality
Detector understanding text descriptions to find matching objects in images
Edge-optimized detector recognizing common objects on resource-limited devices effectively
Lightweight detector for finding rotated objects in aerial imagery on edge hardware
Edge-friendly detector recognizing hundreds of object categories on resource-limited devices effectively
Edge-optimized model producing object outlines on devices with limited resources.
Detector offering fast performance on mid-range graphics cards and processors
Efficiently finds rotated objects in aerial photos for mapping and analysis tasks
Compact detector recognizing diverse object types across many different image categories
Fast model creating object masks for real-time image segmentation needs
Lightweight detector finding objects based on text descriptions for edge applications
High-accuracy detector for critical applications where precision matters most
High-precision detector for rotated objects in aerial and satellite imagery analysis
Accurate general detector recognizing over six hundred different object types
High-accuracy model generating detailed object outlines for demanding professional applications
Open-vocabulary detector with high accuracy for text-based object finding
Detector enhanced with transformer technology for improved object finding capabilities
Compact model producing both masks and boxes with transformer-enhanced accuracy
Advanced detector with transformer backbone delivering superior accuracy for complex scenes
Advanced model creating precise object outlines using enhanced transformer architecture
Finds any object you name in images without requiring training on those specific items
Hugging Face Transformers model for zero-shot object detection
Torch models#
alexnet-imagenet-torch#
Classic neural network that recognizes images and helped launch the deep learning revolution.
Details
Model name:
alexnet-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Alex Krizhevsky
Model license: BSD 3-Clause
Model size: 233.10 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, alexnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("alexnet-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
classification-transformer-torch#
Vision transformer for image classification and custom fine-tuning on specialized datasets.
Details
Model name:
classification-transformer-torch
Model source: https://huggingface.co/docs/transformers/tasks/image_classification
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? yes
Tags:
classification, logits, embeddings, torch, transformers, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("classification-transformer-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
clip-vit-base32-torch#
Understands both images and text together, enabling search and classification using natural language descriptions.
Details
Model name:
clip-vit-base32-torch
Model source: openai/CLIP
Model author: Alec Radford, et al.
Model license: MIT
Model size: 337.58 MB
Exposes embeddings? yes
Tags:
classification, logits, embeddings, torch, clip, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("clip-vit-base32-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "clip-vit-base32-torch",
24 text_prompt="A photo of a",
25 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
26)
27
28dataset.apply_model(model, label_field="predictions")
29session.refresh()
convnext-base-224-torch#
Base modern CNN with transformer elements for robust visual understanding.
Details
Model name:
convnext-base-224-torch
Model source: https://huggingface.co/facebook/convnext-base-224
Model author: Zhuang Liu, et al.
Model license: Apache 2.0
Model size: 1.32 GB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, convnext, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("convnext-base-224-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
convnext-large-224-torch#
Large modern CNN demonstrating competitive performance with vision transformers.
Details
Model name:
convnext-large-224-torch
Model source: https://huggingface.co/facebook/convnext-large-224
Model author: Zhuang Liu, et al.
Model license: Apache 2.0
Model size: 2.95 GB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, convnext
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("convnext-large-224-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
convnext-small-224-torch#
Small modernized CNN delivering strong accuracy through architectural innovations.
Details
Model name:
convnext-small-224-torch
Model source: https://huggingface.co/facebook/convnext-small-224
Model author: Zhuang Liu, et al.
Model license: Apache 2.0
Model size: 383.39 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, convnext, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("convnext-small-224-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
convnext-tiny-224-torch#
Tiny modern CNN bridging traditional convolutions with transformer-inspired improvements.
Details
Model name:
convnext-tiny-224-torch
Model source: https://huggingface.co/facebook/convnext-tiny-224
Model author: Zhuang Liu, et al.
Model license: Apache 2.0
Model size: 436.54 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, convnext
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("convnext-tiny-224-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
convnext-xlarge-224-torch#
Extra-large modern CNN maximizing architectural improvements for top accuracy.
Details
Model name:
convnext-xlarge-224-torch
Model source: https://huggingface.co/facebook/convnext-xlarge-224-22k-1k
Model author: Zhuang Liu, et al.
Model license: Apache 2.0
Model size: 1.30 GB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, convnext, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("convnext-xlarge-224-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
deeplabv3-resnet101-coco-torch#
Labels everyday objects in images pixel by pixel for general scene understanding and analysis.
Details
Model name:
deeplabv3-resnet101-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Liang-Chieh Chen, et al.
Model license: BSD 3-Clause
Model size: 233.22 MB
Exposes embeddings? no
Tags:
segmentation, coco, torch, resnet, deeplabv3, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("deeplabv3-resnet101-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
deeplabv3-resnet50-coco-torch#
Faster version that quickly identifies and labels objects in images for real-time applications.
Details
Model name:
deeplabv3-resnet50-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Liang-Chieh Chen, et al.
Model license: BSD 3-Clause
Model size: 160.51 MB
Exposes embeddings? no
Tags:
segmentation, coco, torch, resnet, deeplabv3, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("deeplabv3-resnet50-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
densenet121-imagenet-torch#
Compact yet powerful classifier that delivers strong results while using minimal computational resources.
Details
Model name:
densenet121-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Gao Huang, et al.
Model license: BSD 3-Clause
Model size: 30.84 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, densenet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("densenet121-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
densenet161-imagenet-torch#
Dense network that achieves high accuracy for image classification and adapts well to new tasks.
Details
Model name:
densenet161-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Gao Huang, et al.
Model license: BSD 3-Clause
Model size: 110.37 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, densenet
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("densenet161-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
densenet169-imagenet-torch#
Deeper variant offering improved accuracy while remaining efficient enough for practical deployment.
Details
Model name:
densenet169-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Gao Huang, et al.
Model license: BSD 3-Clause
Model size: 54.71 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, densenet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("densenet169-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
densenet201-imagenet-torch#
Extra-deep model providing the most detailed features for complex image understanding tasks.
Details
Model name:
densenet201-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Gao Huang, et al.
Model license: BSD 3-Clause
Model size: 77.37 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, densenet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("densenet201-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
depth-estimation-transformer-torch#
Hugging Face Transformers model for monocular depth estimation.
Details
Model name:
depth-estimation-transformer-torch
Model source: https://huggingface.co/docs/transformers/tasks/monocular_depth_estimation
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? no
Tags:
depth, torch, transformers
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("depth-estimation-transformer-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
detection-transformer-torch#
Modern object detector that finds items in images without needing complex post-processing steps.
Details
Model name:
detection-transformer-torch
Model source: https://huggingface.co/docs/transformers/tasks/object_detection
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? yes
Tags:
detection, logits, embeddings, torch, transformers, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("detection-transformer-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
dfine-large-coco-torch#
D-FINE Large from D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement
trained on COCO. Achieves 54.0% AP at 124 FPS on T4 GPU..
Details
Model name:
dfine-large-coco-torch
Model source: Peterande/D-FINE
Model author: Yansong Peng, et al.
Model license: Apache 2.0
Model size: 50.00 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformers, detr, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dfine-large-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
dfine-medium-coco-torch#
D-FINE Medium from D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement
trained on COCO. Mid-size real-time object detector..
Details
Model name:
dfine-medium-coco-torch
Model source: Peterande/D-FINE
Model author: Yansong Peng, et al.
Model license: Apache 2.0
Model size: 30.00 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformers, detr, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dfine-medium-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
dfine-nano-coco-torch#
D-FINE Nano from D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement
trained on COCO. Ultra-lightweight real-time object detector..
Details
Model name:
dfine-nano-coco-torch
Model source: Peterande/D-FINE
Model author: Yansong Peng, et al.
Model license: Apache 2.0
Model size: 14.00 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformers, detr, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dfine-nano-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
dfine-small-coco-torch#
D-FINE Small from D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement
trained on COCO. Balanced real-time object detector..
Details
Model name:
dfine-small-coco-torch
Model source: Peterande/D-FINE
Model author: Yansong Peng, et al.
Model license: Apache 2.0
Model size: 24.00 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformers, detr, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dfine-small-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
dfine-xlarge-coco-torch#
D-FINE XLarge from D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement
trained on COCO. Achieves 55.8% AP at 78 FPS on T4 GPU..
Details
Model name:
dfine-xlarge-coco-torch
Model source: Peterande/D-FINE
Model author: Yansong Peng, et al.
Model license: Apache 2.0
Model size: 123.00 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformers, detr, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dfine-xlarge-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
dinov2-vitb14-reg-torch#
Enhanced image search model that resists noise and errors for more reliable similarity matching.
Details
Model name:
dinov2-vitb14-reg-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 330.35 MB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vitb14-reg-torch")
13
14embeddings = dataset.compute_embeddings(model)
dinov2-vitb14-torch#
Creates searchable image fingerprints for finding similar pictures and organizing large photo collections.
Details
Model name:
dinov2-vitb14-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 330.33 MB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vitb14-torch")
13
14embeddings = dataset.compute_embeddings(model)
dinov2-vitg14-reg-torch#
Highest-capacity dinov2 search model with maximum stability for finding images across massive diverse datasets.
Details
Model name:
dinov2-vitg14-reg-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 4.23 GB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vitg14-reg-torch")
13
14embeddings = dataset.compute_embeddings(model)
dinov2-vitg14-torch#
Powerful image search engine that handles enormous photo collections with rich detail extraction.
Details
Model name:
dinov2-vitg14-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 4.23 GB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vitg14-torch")
13
14embeddings = dataset.compute_embeddings(model)
dinov2-vitl14-reg-torch#
Large stable model for finding and grouping similar images across big databases reliably.
Details
Model name:
dinov2-vitl14-reg-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 1.13 GB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vitl14-reg-torch")
13
14embeddings = dataset.compute_embeddings(model)
dinov2-vitl14-torch#
Large model that creates detailed image fingerprints for advanced search and automatic grouping.
Details
Model name:
dinov2-vitl14-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 1.13 GB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vitl14-torch")
13
14embeddings = dataset.compute_embeddings(model)
dinov2-vits14-reg-torch#
Compact stable model for image search that runs efficiently on phones and edge devices.
Details
Model name:
dinov2-vits14-reg-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 84.20 MB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vits14-reg-torch")
13
14embeddings = dataset.compute_embeddings(model)
dinov2-vits14-torch#
Small model enabling image search and similarity matching directly on mobile devices.
Details
Model name:
dinov2-vits14-torch
Model source: facebookresearch/dinov2
Model author: Maxime Oquab, et al.
Model license: Apache 2.0
Model size: 84.19 MB
Exposes embeddings? yes
Tags:
embeddings, torch, dinov2, transformer, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("dinov2-vits14-torch")
13
14embeddings = dataset.compute_embeddings(model)
efficientnet-b0-imagenet-torch#
Efficient image classifier optimized for mobile devices with excellent accuracy-efficiency tradeoff.
Details
Model name:
efficientnet-b0-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b0
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 81.80 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b0-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
efficientnet-b1-imagenet-torch#
Scaled efficient classifier with improved accuracy for slightly larger computational budgets.
Details
Model name:
efficientnet-b1-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b1
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 120.47 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b1-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
efficientnet-b2-imagenet-torch#
Balanced efficient model providing stronger performance while maintaining reasonable resource usage.
Details
Model name:
efficientnet-b2-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b2
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 140.63 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b2-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
efficientnet-b3-imagenet-torch#
Mid-scale efficient classifier delivering high accuracy for versatile deployment scenarios.
Details
Model name:
efficientnet-b3-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b3
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 188.64 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b3-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
efficientnet-b4-imagenet-torch#
Large efficient model with enhanced features for transfer learning applications.
Details
Model name:
efficientnet-b4-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b4
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 297.81 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b4-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
efficientnet-b5-imagenet-torch#
High-capacity efficient classifier prioritizing accuracy with available compute resources.
Details
Model name:
efficientnet-b5-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b5
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 467.25 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b5-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
efficientnet-b6-imagenet-torch#
Extended efficient model approaching state-of-the-art accuracy on challenging datasets.
Details
Model name:
efficientnet-b6-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b6
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 661.19 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b6-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
efficientnet-b7-imagenet-torch#
Maximum efficient classifier pushing performance boundaries while preserving efficiency principles.
Details
Model name:
efficientnet-b7-imagenet-torch
Model source: https://huggingface.co/google/efficientnet-b7
Model author: Mingxing Tan and Quoc V. Le
Model license: Apache 2.0
Model size: 1018.36 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, efficientnet, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("efficientnet-b7-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
faster-rcnn-resnet50-fpn-coco-torch#
Multi-scale object finder that accurately detects both small and large items in images.
Details
Model name:
faster-rcnn-resnet50-fpn-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Shaoqing Ren, et al.
Model license: BSD 3-Clause
Model size: 159.74 MB
Exposes embeddings? no
Tags:
detection, coco, torch, faster-rcnn, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-resnet50-fpn-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
fcn-resnet101-coco-torch#
Creates detailed pixel-level labels for images, identifying and outlining twenty-one different object categories.
Details
Model name:
fcn-resnet101-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Jonathan Long, et al.
Model license: BSD 3-Clause
Model size: 207.71 MB
Exposes embeddings? no
Tags:
segmentation, coco, torch, fcn, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("fcn-resnet101-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
fcn-resnet50-coco-torch#
Fast image labeler that quickly identifies and outlines objects for interactive editing and annotation.
Details
Model name:
fcn-resnet50-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Jonathan Long, et al.
Model license: BSD 3-Clause
Model size: 135.01 MB
Exposes embeddings? no
Tags:
segmentation, coco, torch, fcn, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("fcn-resnet50-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
googlenet-imagenet-torch#
Classic image classifier providing reliable categorization and features for various computer vision projects..
Details
Model name:
googlenet-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Christian Szegedy, et al.
Model license: BSD 3-Clause
Model size: 49.73 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, googlenet, official
Requirements
Packages:
scipy, torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("googlenet-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
group-vit-segmentation-transformer-torch#
Hugging Face Transformers model for zero-shot semantic segmentation.
Details
Model name:
group-vit-segmentation-transformer-torch
Model source: https://huggingface.co/docs/transformers/en/tasks/mask_generation
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Model size: 212.80 MB
Exposes embeddings? yes
Tags:
segmentation, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("group-vit-segmentation-transformer-torch",
13 text_prompt="A photo of a",
14 classes=["person", "dog", "cat", "bird", "car", "tree", "other"])
15
16dataset.apply_model(model, label_field="predictions")
17
18session = fo.launch_app(dataset)
inception-v3-imagenet-torch#
Efficient image classifier delivering accurate results with useful features for transfer learning applications.
Details
Model name:
inception-v3-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Christian Szegedy, et al.
Model license: BSD 3-Clause
Model size: 103.81 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, inception, official
Requirements
Packages:
scipy, torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("inception-v3-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
keypoint-rcnn-resnet50-fpn-coco-torch#
Finds people in images and maps their body joints for pose estimation and motion analysis.
Details
Model name:
keypoint-rcnn-resnet50-fpn-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Kaiming He, et al.
Model license: BSD 3-Clause
Model size: 226.05 MB
Exposes embeddings? no
Tags:
keypoints, coco, torch, keypoint-rcnn, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("keypoint-rcnn-resnet50-fpn-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
mask-rcnn-resnet50-fpn-coco-torch#
Multi-scale object outliner using advanced architecture for accurate segmentation across different object sizes.
Details
Model name:
mask-rcnn-resnet50-fpn-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Kaiming He, et al.
Model license: BSD 3-Clause
Model size: 169.84 MB
Exposes embeddings? no
Tags:
instances, coco, torch, mask-rcnn, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("mask-rcnn-resnet50-fpn-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
med-sam-2-video-torch#
Medical segmentation tool that outlines organs and structures in medical videos and 3D scans.
Details
Model name:
med-sam-2-video-torch
Model source: MedicineToken/Medical-SAM2
Model author: Jiayuan Zhu, et al.
Model license: Apache 2.0
Model size: 74.46 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, med-SAM, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4from fiftyone.utils.huggingface import load_from_hub
5
6dataset = load_from_hub("Voxel51/BTCV-CT-as-video-MedSAM2-dataset")[:2]
7
8# Retaining detections from a single frame in the middle
9# Note that SAM2 only propagates segmentation masks forward in a video
10(
11 dataset
12 .match_frames(F("frame_number") != 100)
13 .set_field("frames.gt_detections", None)
14 .save()
15)
16
17model = foz.load_zoo_model("med-sam-2-video-torch")
18
19# Segment inside boxes and propagate to all frames
20dataset.apply_model(
21 model,
22 label_field="pred_segmentations",
23 prompt_field="frames.gt_detections",
24)
25
26session = fo.launch_app(dataset)
mnasnet0.5-imagenet-torch#
Ultra-lightweight image classifier designed by AI for running directly on phones and IoT devices.
Details
Model name:
mnasnet0.5-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Mingxing Tan, et al.
Model license: BSD 3-Clause
Model size: 8.59 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, mnasnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("mnasnet0.5-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
mnasnet1.0-imagenet-torch#
Mobile-optimized classifier balancing size and accuracy for efficient on-device image recognition.
Details
Model name:
mnasnet1.0-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Mingxing Tan, et al.
Model license: BSD 3-Clause
Model size: 16.92 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, mnasnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("mnasnet1.0-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
mobilenet-v2-imagenet-torch#
Mobile-friendly image classifier optimized for quick training and deployment on resource-limited devices.
Details
Model name:
mobilenet-v2-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Mark Sandler, et al.
Model license: BSD 3-Clause
Model size: 13.55 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, mobilenet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("mobilenet-v2-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
omdet-turbo-swin-tiny-torch#
Real-time detector that finds any object you describe in words, perfect for live video analysis.
Details
Model name:
omdet-turbo-swin-tiny-torch
Model source: https://huggingface.co/docs/transformers/tasks/zero_shot_object_detection
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? yes
Tags:
detection, logits, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers>=4.51
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12
13classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]
14
15model = foz.load_zoo_model(
16 "omdet-turbo-swin-tiny-torch",
17 classes=classes,
18)
19
20dataset.apply_model(model, label_field="predictions")
21
22session = fo.launch_app(dataset)
open-clip-torch#
Connects images with text descriptions enabling search by words and automatic content filtering systems.
Details
Model name:
open-clip-torch
Model source: mlfoundations/open_clip
Model author: Gabriel Ilharco, et al.
Model license: MIT
Exposes embeddings? yes
Tags:
classification, logits, embeddings, torch, clip, zero-shot, transformer
Requirements
Packages:
torch, torchvision, open_clip_torch
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("open-clip-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "open-clip-torch",
24 text_prompt="A photo of a",
25 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
26)
27
28dataset.apply_model(model, label_field="predictions")
29session.refresh()
owlvit-base-patch16-torch#
Finds any object you name in pictures using 16x16 image patches without needing specific training for those items.
Details
Model name:
owlvit-base-patch16-torch
Model source: https://huggingface.co/docs/transformers/tasks/zero_shot_object_detection
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? yes
Tags:
detection, logits, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12
13classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]
14
15model = foz.load_zoo_model(
16 "owlvit-base-patch16-torch",
17 classes=classes,
18)
19
20dataset.apply_model(model, label_field="predictions")
21
22session = fo.launch_app(dataset)
owlvit-base-patch32-torch#
Finds any object you name in pictures using efficient 32x32 image patches without needing specific training.
Details
Model name:
owlvit-base-patch32-torch
Model source: https://huggingface.co/docs/transformers/tasks/zero_shot_object_detection
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Model size: 1.14 GB
Exposes embeddings? yes
Tags:
detection, logits, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12
13classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]
14
15model = foz.load_zoo_model(
16 "owlvit-base-patch32-torch",
17 classes=classes,
18)
19
20dataset.apply_model(model, label_field="predictions")
21
22session = fo.launch_app(dataset)
owlvit-large-patch14-torch#
Large OWL-ViT zero-shot object detector with ViT-L/14 backbone. Achieves higher accuracy than base models, especially for smaller objects..
Details
Model name:
owlvit-large-patch14-torch
Model source: https://huggingface.co/google/owlvit-large-patch14
Model author: Matthias Minderer, et al.
Model license: Apache 2.0
Model size: 1.62 GB
Exposes embeddings? yes
Tags:
detection, logits, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12
13classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]
14
15model = foz.load_zoo_model(
16 "owlvit-large-patch14-torch",
17 classes=classes,
18)
19
20dataset.apply_model(model, label_field="predictions")
21
22session = fo.launch_app(dataset)
resnet101-imagenet-torch#
Deep image recognition model delivering high accuracy for demanding classification and analysis tasks.
Details
Model name:
resnet101-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Kaiming He, et al.
Model license: BSD 3-Clause
Model size: 170.45 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnet101-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnet152-imagenet-torch#
Very deep classifier providing the richest visual features for precision-critical image understanding applications.
Details
Model name:
resnet152-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Kaiming He, et al.
Model license: BSD 3-Clause
Model size: 230.34 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnet152-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnet18-imagenet-torch#
Lightweight image classifier designed for fast recognition on phones and other resource-limited devices.
Details
Model name:
resnet18-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Kaiming He, et al.
Model license: BSD 3-Clause
Model size: 44.66 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnet18-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnet34-imagenet-torch#
Balanced image classifier offering good accuracy and speed for everyday computer vision needs.
Details
Model name:
resnet34-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Kaiming He, et al.
Model license: BSD 3-Clause
Model size: 83.26 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnet34-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnet50-imagenet-torch#
Most popular image recognition backbone widely used as starting point for custom vision projects.
Details
Model name:
resnet50-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Kaiming He, et al.
Model license: BSD 3-Clause
Model size: 97.75 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnet50-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnext101-32x8d-imagenet-torch#
Powerful image classifier with enhanced capacity for handling complex visual recognition challenges effectively.
Details
Model name:
resnext101-32x8d-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Saining Xie, et al.
Model license: BSD 3-Clause
Model size: 339.59 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, resnext, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnext101-32x8d-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnext50-32x4d-imagenet-torch#
Efficient advanced classifier delivering strong accuracy with reasonable computing requirements for practical deployments.
Details
Model name:
resnext50-32x4d-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Saining Xie, et al.
Model license: BSD 3-Clause
Model size: 95.79 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, resnext
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnext50-32x4d-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
retinanet-resnet50-fpn-coco-torch#
Fast object detector that quickly finds and boxes eighty common items in any image.
Details
Model name:
retinanet-resnet50-fpn-coco-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Tsung-Yi Lin, et al.
Model license: BSD 3-Clause
Model size: 130.27 MB
Exposes embeddings? no
Tags:
detection, coco, torch, retinanet, resnet
Requirements
Packages:
torch, torchvision>=0.8.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("retinanet-resnet50-fpn-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
rtdetr-l-coco-torch#
Modern real-time object detector that finds items without complex post-processing for responsive applications.
Details
Model name:
rtdetr-l-coco-torch
Model source: https://docs.ultralytics.com/models/rtdetr/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 63.43 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformer, rtdetr, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.2.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("rtdetr-l-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
rtdetr-v2-m-coco-torch#
Balanced real-time object detector offering improved accuracy for production use.
Details
Model name:
rtdetr-v2-m-coco-torch
Model source: https://huggingface.co/PekingU/rtdetr_v2_r50vd
Model author: Wenyu Lv, et al.
Model license: Apache-2.0
Model size: 328.41 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformers, rtdetr, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("rtdetr-v2-m-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
rtdetr-v2-s-coco-torch#
Lightweight real-time object detector optimized for speed on edge devices.
Details
Model name:
rtdetr-v2-s-coco-torch
Model source: https://huggingface.co/PekingU/rtdetr_v2_r18vd
Model author: Wenyu Lv, et al.
Model license: Apache-2.0
Model size: 154.32 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformers, rtdetr
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("rtdetr-v2-s-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
rtdetr-x-coco-torch#
High-capacity object detector delivering very precise results at speeds suitable for production use.
Details
Model name:
rtdetr-x-coco-torch
Model source: https://docs.ultralytics.com/models/rtdetr/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 129.47 MB
Exposes embeddings? no
Tags:
detection, coco, torch, transformer, rtdetr, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.2.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("rtdetr-x-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
segformer-b0-ade20k-torch#
Efficient transformer-based semantic segmentation model for scene parsing with 150 classes.
Details
Model name:
segformer-b0-ade20k-torch
Model source: https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512
Model author: Enze Xie, et al.
Model license: MIT
Model size: 14.32 MB
Exposes embeddings? no
Tags:
segmentation, torch, segformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segformer-b0-ade20k-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
segformer-b1-ade20k-torch#
Balanced SegFormer model providing good accuracy-efficiency tradeoff for scene understanding.
Details
Model name:
segformer-b1-ade20k-torch
Model source: https://huggingface.co/nvidia/segformer-b1-finetuned-ade-512-512
Model author: Enze Xie, et al.
Model license: MIT
Model size: 52.32 MB
Exposes embeddings? no
Tags:
segmentation, torch, segformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segformer-b1-ade20k-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
segformer-b2-ade20k-torch#
Medium-sized SegFormer delivering enhanced segmentation quality for complex scenes.
Details
Model name:
segformer-b2-ade20k-torch
Model source: https://huggingface.co/nvidia/segformer-b2-finetuned-ade-512-512
Model author: Enze Xie, et al.
Model license: MIT
Model size: 104.76 MB
Exposes embeddings? no
Tags:
segmentation, torch, segformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segformer-b2-ade20k-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
segformer-b3-ade20k-torch#
Larger SegFormer model with improved accuracy for detailed semantic segmentation.
Details
Model name:
segformer-b3-ade20k-torch
Model source: https://huggingface.co/nvidia/segformer-b3-finetuned-ade-512-512
Model author: Enze Xie, et al.
Model license: MIT
Model size: 180.59 MB
Exposes embeddings? no
Tags:
segmentation, torch, segformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segformer-b3-ade20k-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
segformer-b4-ade20k-torch#
High-capacity SegFormer achieving excellent results on challenging segmentation tasks.
Details
Model name:
segformer-b4-ade20k-torch
Model source: https://huggingface.co/nvidia/segformer-b4-finetuned-ade-512-512
Model author: Enze Xie, et al.
Model license: MIT
Model size: 244.56 MB
Exposes embeddings? no
Tags:
segmentation, torch, segformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segformer-b4-ade20k-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
segformer-b5-ade20k-torch#
Largest SegFormer model delivering the best semantic segmentation performance in its family.
Details
Model name:
segformer-b5-ade20k-torch
Model source: https://huggingface.co/nvidia/segformer-b5-finetuned-ade-640-640
Model author: Enze Xie, et al.
Model license: MIT
Model size: 323.14 MB
Exposes embeddings? no
Tags:
segmentation, torch, segformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segformer-b5-ade20k-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
segment-anything-2-hiera-base-plus-image-torch#
Accurate image segmentation model for editing, labeling, and creative work with still pictures.
Details
Model name:
segment-anything-2-hiera-base-plus-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 308.51 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2-hiera-base-plus-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2-hiera-base-plus-video-torch#
Video segmentation model that tracks and outlines objects throughout clips for editing and analysis.
Details
Model name:
segment-anything-2-hiera-base-plus-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 308.51 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2-hiera-base-plus-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-2-hiera-large-image-torch#
High-quality image segmenter producing detailed masks for demanding professional editing and annotation tasks.
Details
Model name:
segment-anything-2-hiera-large-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 856.35 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2-hiera-large-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2-hiera-large-video-torch#
Advanced video segmenter providing fine object tracking throughout full videos for post-production work.
Details
Model name:
segment-anything-2-hiera-large-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 856.35 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2-hiera-large-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-2-hiera-small-image-torch#
Fast image segmentation model that runs efficiently on laptops and edge computing devices.
Details
Model name:
segment-anything-2-hiera-small-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 175.77 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2-hiera-small-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2-hiera-small-video-torch#
Quick video segmentation model delivering rapid object tracking on standard graphics cards.
Details
Model name:
segment-anything-2-hiera-small-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 175.77 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2-hiera-small-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-2-hiera-tiny-image-torch#
Smallest image segmentation model offering instant results for mobile apps and embedded systems.
Details
Model name:
segment-anything-2-hiera-tiny-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 148.68 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2-hiera-tiny-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2-hiera-tiny-video-torch#
Tiny video segmenter enabling real-time object tracking on phones and compact devices.
Details
Model name:
segment-anything-2-hiera-tiny-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 148.68 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2-hiera-tiny-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-base-plus-image-torch#
Updated image segmenter with improved mask accuracy for everyday editing and dataset creation.
Details
Model name:
segment-anything-2.1-hiera-base-plus-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 308.62 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2.1-hiera-base-plus-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-base-plus-video-torch#
Enhanced video segmenter with better tracking quality for video analysis and scene understanding.
Details
Model name:
segment-anything-2.1-hiera-base-plus-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 308.62 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2.1-hiera-base-plus-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-large-image-torch#
Large updated model offering even finer masks for high-resolution professional image workflows.
Details
Model name:
segment-anything-2.1-hiera-large-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 856.48 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2.1-hiera-large-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-large-video-torch#
Large video model producing exceptionally detailed masks throughout long videos for intensive production.
Details
Model name:
segment-anything-2.1-hiera-large-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 856.48 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2.1-hiera-large-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-small-image-torch#
Balanced updated segmenter combining speed and accuracy for edge device image processing.
Details
Model name:
segment-anything-2.1-hiera-small-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 175.87 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2.1-hiera-small-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-small-video-torch#
Improved video segmenter maintaining quick performance on compact hardware while enhancing mask quality.
Details
Model name:
segment-anything-2.1-hiera-small-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 175.87 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer, official
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2.1-hiera-small-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-tiny-image-torch#
Enhanced mobile image segmenter for apps, augmented reality filters, and on-device processing.
Details
Model name:
segment-anything-2.1-hiera-tiny-image-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 148.68 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, transformer
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-2.1-hiera-tiny-image-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-2.1-hiera-tiny-video-torch#
Upgraded mobile video segmenter for live effects on phones, wearables, and smart cameras.
Details
Model name:
segment-anything-2.1-hiera-tiny-video-torch
Model source: https://ai.meta.com/sam2/
Model author: Nikhila Ravi, et al.
Model license: Apache 2.0, BSD 3-Clause
Model size: 148.68 MB
Exposes embeddings? no
Tags:
segment-anything, torch, zero-shot, video, transformer
Requirements
Packages:
torch, torchvision, sam2
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3from fiftyone import ViewField as F
4
5dataset = foz.load_zoo_dataset("quickstart-video", max_samples=2)
6
7# Only retain detections in the first frame
8(
9 dataset
10 .match_frames(F("frame_number") > 1)
11 .set_field("frames.detections", None)
12 .save()
13)
14
15model = foz.load_zoo_model("segment-anything-2.1-hiera-tiny-video-torch")
16
17# Segment inside boxes and propagate to all frames
18dataset.apply_model(
19 model,
20 label_field="segmentations",
21 prompt_field="frames.detections", # can contain Detections or Keypoints
22)
23
24session = fo.launch_app(dataset)
segment-anything-vitb-torch#
Interactive segmentation tool that instantly outlines any object you point to or describe.
Details
Model name:
segment-anything-vitb-torch
Model source: https://segment-anything.com
Model author: Alexander Kirillov, et al.
Model license: Apache 2.0
Model size: 357.67 MB
Exposes embeddings? no
Tags:
segment-anything, sa-1b, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, segment-anything
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-vitb-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-vith-torch#
Highest quality segmentation model creating extremely detailed masks for research and large-scale annotation projects.
Details
Model name:
segment-anything-vith-torch
Model source: https://segment-anything.com
Model author: Alexander Kirillov, et al.
Model license: Apache 2.0
Model size: 2.39 GB
Exposes embeddings? no
Tags:
segment-anything, sa-1b, torch, zero-shot, transformer, official
Requirements
Packages:
torch, torchvision, segment-anything
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-vith-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segment-anything-vitl-torch#
Large segmentation model producing finer object outlines for professional editing and labeling workflows.
Details
Model name:
segment-anything-vitl-torch
Model source: https://segment-anything.com
Model author: Alexander Kirillov, et al.
Model license: Apache 2.0
Model size: 1.16 GB
Exposes embeddings? no
Tags:
segment-anything, sa-1b, torch, zero-shot, transformer
Requirements
Packages:
torch, torchvision, segment-anything
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segment-anything-vitl-torch")
13
14# Segment inside boxes
15dataset.apply_model(
16 model,
17 label_field="segmentations",
18 prompt_field="ground_truth", # can contain Detections or Keypoints
19)
20
21# Full automatic segmentations
22dataset.apply_model(model, label_field="auto")
23
24session = fo.launch_app(dataset)
segmentation-transformer-torch#
Hugging Face Transformers model for semantic segmentation.
Details
Model name:
segmentation-transformer-torch
Model source: https://huggingface.co/docs/transformers/tasks/semantic_segmentation
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? no
Tags:
segmentation, torch, transformers, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("segmentation-transformer-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
shufflenetv2-0.5x-imagenet-torch#
Ultra-small image classifier for tiny devices with very limited power and memory.
Details
Model name:
shufflenetv2-0.5x-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Ningning Ma, et al.
Model license: BSD 3-Clause
Model size: 5.28 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, shufflenet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("shufflenetv2-0.5x-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
shufflenetv2-1.0x-imagenet-torch#
Mobile image classifier that works efficiently on phones with modest computing resources.
Details
Model name:
shufflenetv2-1.0x-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Ningning Ma, et al.
Model license: BSD 3-Clause
Model size: 8.79 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, shufflenet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("shufflenetv2-1.0x-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
siglip-base-patch16-224-torch#
Hugging Face Transformers model for zero-shot image classification.
Details
Model name:
siglip-base-patch16-224-torch
Model source: https://huggingface.co/docs/transformers/tasks/zero_shot_image_classification
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Model size: 775.11 MB
Exposes embeddings? yes
Tags:
classification, logits, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers>=4.51
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12
13classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]
14
15model = foz.load_zoo_model(
16 "siglip-base-patch16-224-torch",
17 classes=classes,
18)
19
20dataset.apply_model(model, label_field="predictions")
21
22session = fo.launch_app(dataset)
squeezenet-1.1-imagenet-torch#
Tiny image classifier that fits in just five megabytes for embedded devices.
Details
Model name:
squeezenet-1.1-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Forrest Iandola
Model license: BSD 2-Clause
Model size: 4.74 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, squeezenet
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("squeezenet-1.1-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
squeezenet-imagenet-torch#
Ultra-compact image classifier perfect for severely resource-constrained hardware and applications.
Details
Model name:
squeezenet-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Forrest Iandola
Model license: BSD 2-Clause
Model size: 4.79 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, squeezenet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("squeezenet-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
swin-v2-base-torch#
Base hierarchical transformer delivering strong results across vision tasks.
Details
Model name:
swin-v2-base-torch
Model source: https://huggingface.co/microsoft/swinv2-base-patch4-window8-256
Model author: Ze Liu, et al.
Model license: MIT
Model size: 1.31 GB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, swin-transformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("swin-v2-base-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
swin-v2-large-torch#
Large hierarchical transformer with enhanced capacity for demanding applications.
Details
Model name:
swin-v2-large-torch
Model source: https://huggingface.co/microsoft/swinv2-large-patch4-window12to16-192to256-22kto1k-ft
Model author: Ze Liu, et al.
Model license: MIT
Model size: 2.93 GB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, swin-transformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("swin-v2-large-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
swin-v2-small-torch#
Small hierarchical transformer balancing efficiency and performance for practical use.
Details
Model name:
swin-v2-small-torch
Model source: https://huggingface.co/microsoft/swinv2-small-patch4-window8-256
Model author: Ze Liu, et al.
Model license: MIT
Model size: 759.35 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, swin-transformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("swin-v2-small-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
swin-v2-tiny-torch#
Tiny hierarchical transformer for efficient visual recognition on edge devices.
Details
Model name:
swin-v2-tiny-torch
Model source: https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256
Model author: Ze Liu, et al.
Model license: MIT
Model size: 432.89 MB
Exposes embeddings? no
Tags:
classification, imagenet, torch, transformers, swin-transformer, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("swin-v2-tiny-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
vgg11-bn-imagenet-torch#
Classic image classifier with stable training useful for various computer vision projects.
Details
Model name:
vgg11-bn-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 506.88 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg11-bn-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vgg11-imagenet-torch#
Simple baseline image classifier valuable for research experimentation and learning purposes.
Details
Model name:
vgg11-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 506.84 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg11-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vgg13-bn-imagenet-torch#
Deeper classic classifier providing stable training process and solid accuracy results overall.
Details
Model name:
vgg13-bn-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 507.59 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg13-bn-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vgg13-imagenet-torch#
Straightforward image classifier valued for easy experimentation and model compression studies.
Details
Model name:
vgg13-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 507.54 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg13-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vgg16-bn-imagenet-torch#
Popular feature extractor widely used for detection, style transfer, and medical imaging.
Details
Model name:
vgg16-bn-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 527.87 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg16-bn-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vgg16-imagenet-torch#
PyTorch version of the popular classifier ready for modern deep learning workflows.
Details
Model name:
vgg16-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 527.80 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg16-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vgg19-bn-imagenet-torch#
Deep classic model providing rich features for style transfer and interpretability analysis.
Details
Model name:
vgg19-bn-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 548.14 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg19-bn-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vgg19-imagenet-torch#
Deep image classifier delivering detailed features for creative applications and research projects.
Details
Model name:
vgg19-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Karen Simonyan, et al.
Model license: BSD 3-Clause
Model size: 548.05 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, vgg, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg19-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
vit-base-patch16-224-imagenet-torch#
Modern image classifier that recognizes objects and provides useful features for various computer vision tasks.
Details
Model name:
vit-base-patch16-224-imagenet-torch
Model source: https://huggingface.co/docs/transformers/tasks/image_classification
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Model size: 330.31 MB
Exposes embeddings? yes
Tags:
classification, logits, embeddings, torch, transformers, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vit-base-patch16-224-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
wide-resnet101-2-imagenet-torch#
Extra-wide deep classifier for high-precision image recognition and advanced transfer learning tasks.
Details
Model name:
wide-resnet101-2-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Sergey Zagoruyko, et al.
Model license: BSD 3-Clause
Model size: 242.90 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, wide-resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("wide-resnet101-2-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
wide-resnet50-2-imagenet-torch#
Wide classifier offering stronger accuracy and better features for adapting to new tasks.
Details
Model name:
wide-resnet50-2-imagenet-torch
Model source: https://pytorch.org/vision/main/models.html
Model author: Sergey Zagoruyko, et al.
Model license: BSD 3-Clause
Model size: 131.82 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, torch, wide-resnet, official
Requirements
Packages:
torch, torchvision
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("wide-resnet50-2-imagenet-torch")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
yolo-nas-torch#
AI-designed detector family offering three model variants for diverse deployment scenarios.
Details
Model name:
yolo-nas-torch
Model source: Deci-AI/super-gradients
Model author: Shay Aharon, et al.
Model license: Apache 2.0
Exposes embeddings? no
Tags:
detection, torch, yolo, official
Requirements
Packages:
torch, torchvision, super-gradients
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo-nas-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11l-coco-torch#
Real-time object detector balancing high accuracy with fast processing speeds effectively.
Details
Model name:
yolo11l-coco-torch
Model source: https://docs.ultralytics.com/models/yolov11/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 49.01 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11l-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11l-seg-coco-torch#
Model creating detailed object outlines for precise image editing and analysis.
Details
Model name:
yolo11l-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolo11/#__tabbed_1_2
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 53.50 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11l-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11m-coco-torch#
Object detector offering good balance between speed and accuracy for most applications.
Details
Model name:
yolo11m-coco-torch
Model source: https://docs.ultralytics.com/models/yolov11/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 38.80 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11m-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11m-seg-coco-torch#
Model generating object masks efficiently for everyday segmentation tasks.
Details
Model name:
yolo11m-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolo11/#__tabbed_1_2
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 43.30 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11m-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11n-coco-torch#
Object detector designed specifically for phones and other edge computing devices.
Details
Model name:
yolo11n-coco-torch
Model source: https://docs.ultralytics.com/models/yolov11/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 5.35 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11n-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11n-seg-coco-torch#
Edge model producing object outlines directly on phones and edge devices.
Details
Model name:
yolo11n-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolo11/#__tabbed_1_2
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 5.90 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11n-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11s-coco-torch#
Fast object detector ideal for systems with limited graphics processing power.
Details
Model name:
yolo11s-coco-torch
Model source: https://docs.ultralytics.com/models/yolov11/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 18.42 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11s-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11s-seg-coco-torch#
Model creating object masks quickly for real-time segmentation applications.
Details
Model name:
yolo11s-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolo11/#__tabbed_1_2
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 19.71 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11s-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11x-coco-torch#
Object detector prioritizing accuracy over processing speed for critical applications.
Details
Model name:
yolo11x-coco-torch
Model source: https://docs.ultralytics.com/models/yolov11/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 109.33 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11x-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolo11x-seg-coco-torch#
Model delivering high-quality object outlines for professional workflows.
Details
Model name:
yolo11x-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolo11/#__tabbed_1_2
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 119.30 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo11x-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yoloe11l-seg-torch#
Real-time model creating both object outlines and boxes for any described item.
Details
Model name:
yoloe11l-seg-torch
Model source: https://docs.ultralytics.com/models/yoloe
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 67.69 MB
Exposes embeddings? no
Tags:
instances, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.99
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yoloe11l-seg-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yoloe11l-seg-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yoloe11m-seg-torch#
Model producing masks and boxes for objects described in natural language.
Details
Model name:
yoloe11m-seg-torch
Model source: https://docs.ultralytics.com/models/yoloe
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 57.48 MB
Exposes embeddings? no
Tags:
instances, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.99
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yoloe11m-seg-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yoloe11m-seg-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yoloe11s-seg-torch#
Segments specified classes, generating object outlines and boxes for real-time applications.
Details
Model name:
yoloe11s-seg-torch
Model source: https://docs.ultralytics.com/models/yoloe
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 26.52 MB
Exposes embeddings? no
Tags:
instances, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.99
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yoloe11s-seg-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yoloe11s-seg-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yoloev8l-seg-torch#
Model outlining and boxing any object you describe without specific training.
Details
Model name:
yoloev8l-seg-torch
Model source: https://docs.ultralytics.com/models/yoloe
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 102.43 MB
Exposes embeddings? no
Tags:
instances, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.99
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yoloev8l-seg-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yoloev8l-seg-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yoloev8m-seg-torch#
Model creating masks for any object type you name in text.
Details
Model name:
yoloev8m-seg-torch
Model source: https://docs.ultralytics.com/models/yoloe
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 62.75 MB
Exposes embeddings? no
Tags:
instances, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.99
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yoloev8m-seg-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yoloev8m-seg-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yoloev8s-seg-torch#
Compact model producing outlines for objects described in words on edge devices.
Details
Model name:
yoloev8s-seg-torch
Model source: https://docs.ultralytics.com/models/yoloe
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 29.69 MB
Exposes embeddings? no
Tags:
instances, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.3.99
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yoloev8s-seg-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yoloev8s-seg-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yolov10l-coco-torch#
Object detector with special optimizations for even faster inference on modern hardware.
Details
Model name:
yolov10l-coco-torch
Model source: https://docs.ultralytics.com/models/yolov10/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 50.00 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.2.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov10l-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov10m-coco-torch#
Balanced detector providing good accuracy and speed for general-purpose object detection tasks.
Details
Model name:
yolov10m-coco-torch
Model source: https://docs.ultralytics.com/models/yolov10/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 32.09 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.2.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov10m-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov10n-coco-torch#
Edge-optimized detector for devices with minimal computing resources available.
Details
Model name:
yolov10n-coco-torch
Model source: https://docs.ultralytics.com/models/yolov10/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 5.59 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.2.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov10n-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov10s-coco-torch#
Fast lightweight detector suitable for systems with limited GPU capabilities and memory.
Details
Model name:
yolov10s-coco-torch
Model source: https://docs.ultralytics.com/models/yolov10/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 15.85 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.2.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov10s-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov10x-coco-torch#
High-accuracy detector for demanding object detection applications and research.
Details
Model name:
yolov10x-coco-torch
Model source: https://docs.ultralytics.com/models/yolov10/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 61.41 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.2.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov10x-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov5l-coco-torch#
Real-time detector producing accurate results quickly for demanding vision applications.
Details
Model name:
yolov5l-coco-torch
Model source: https://pytorch.org/hub/ultralytics_yolov5
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 101.96 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov5l-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov5m-coco-torch#
Real-time detector balancing good accuracy with fast processing speeds.
Details
Model name:
yolov5m-coco-torch
Model source: https://pytorch.org/hub/ultralytics_yolov5
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 48.25 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov5m-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov5n-coco-torch#
Lightweight detector for edge devices needing basic object detection capabilities.
Details
Model name:
yolov5n-coco-torch
Model source: https://pytorch.org/hub/ultralytics_yolov5
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 5.31 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov5n-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov5s-coco-torch#
Real-time detector delivering good results with minimal computational requirements.
Details
Model name:
yolov5s-coco-torch
Model source: https://pytorch.org/hub/ultralytics_yolov5
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 17.72 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov5s-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov5x-coco-torch#
High-accuracy detector offering top precision for applications where quality is critical.
Details
Model name:
yolov5x-coco-torch
Model source: https://pytorch.org/hub/ultralytics_yolov5
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 186.09 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov5x-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8l-coco-torch#
Real-time detector with advanced architecture for improved object finding in complex scenes.
Details
Model name:
yolov8l-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 83.70 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8l-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8l-obb-dotav1-torch#
Specialized detector that finds rotated objects in aerial and satellite imagery accurately.
Details
Model name:
yolov8l-obb-dotav1-torch
Model source: https://docs.ultralytics.com/tasks/obb/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 85.36 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, polylines, obb, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8l-obb-dotav1-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8l-oiv7-torch#
General-purpose detector trained on diverse images recognizing over six hundred object categories.
Details
Model name:
yolov8l-oiv7-torch
Model source: https://docs.ultralytics.com/datasets/detect/open-images-v7
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 83.70 MB
Exposes embeddings? no
Tags:
detection, oiv7, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8l-oiv7-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8l-seg-coco-torch#
Creates precise object outlines for detailed image editing and analysis tasks.
Details
Model name:
yolov8l-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 88.11 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8l-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8l-world-torch#
Finds and boxes any object you describe using natural language prompts.
Details
Model name:
yolov8l-world-torch
Model source: https://docs.ultralytics.com/models/yolo-world/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 91.23 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8l-world-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yolov8l-world-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yolov8m-coco-torch#
Detector balancing speed and accuracy for everyday object detection needs.
Details
Model name:
yolov8m-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 49.70 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8m-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8m-obb-dotav1-torch#
Finds rotated bounding boxes in aerial images for mapping and surveillance applications.
Details
Model name:
yolov8m-obb-dotav1-torch
Model source: https://docs.ultralytics.com/tasks/obb/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 50.84 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, polylines, obb, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8m-obb-dotav1-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8m-oiv7-torch#
Versatile detector recognizing hundreds of different object types across varied image domains.
Details
Model name:
yolov8m-oiv7-torch
Model source: https://docs.ultralytics.com/datasets/detect/open-images-v7
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 50.29 MB
Exposes embeddings? no
Tags:
detection, oiv7, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8m-oiv7-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8m-seg-coco-torch#
Generates object masks with good balance of speed and quality.
Details
Model name:
yolov8m-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 52.36 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8m-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8m-world-torch#
Detector understanding text descriptions to find matching objects in images.
Details
Model name:
yolov8m-world-torch
Model source: https://docs.ultralytics.com/models/yolo-world/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 55.89 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8m-world-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yolov8m-world-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yolov8n-coco-torch#
Edge-optimized detector recognizing common objects on resource-limited devices effectively.
Details
Model name:
yolov8n-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 6.23 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8n-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8n-obb-dotav1-torch#
Lightweight detector for finding rotated objects in aerial imagery on edge hardware.
Details
Model name:
yolov8n-obb-dotav1-torch
Model source: https://docs.ultralytics.com/tasks/obb/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 6.24 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, polylines, obb, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8n-obb-dotav1-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8n-oiv7-torch#
Edge-friendly detector recognizing hundreds of object categories on resource-limited devices effectively.
Details
Model name:
yolov8n-oiv7-torch
Model source: https://docs.ultralytics.com/datasets/detect/open-images-v7
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 6.89 MB
Exposes embeddings? no
Tags:
detection, oiv7, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8n-oiv7-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8n-seg-coco-torch#
Edge-optimized model producing object outlines on devices with limited resources..
Details
Model name:
yolov8n-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 6.73 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8n-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8s-coco-torch#
Detector offering fast performance on mid-range graphics cards and processors.
Details
Model name:
yolov8s-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 21.53 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8s-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8s-obb-dotav1-torch#
Efficiently finds rotated objects in aerial photos for mapping and analysis tasks.
Details
Model name:
yolov8s-obb-dotav1-torch
Model source: https://docs.ultralytics.com/tasks/obb/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 22.17 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, polylines, obb, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8s-obb-dotav1-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8s-oiv7-torch#
Compact detector recognizing diverse object types across many different image categories.
Details
Model name:
yolov8s-oiv7-torch
Model source: https://docs.ultralytics.com/datasets/detect/open-images-v7
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 21.92 MB
Exposes embeddings? no
Tags:
detection, oiv7, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8s-oiv7-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8s-seg-coco-torch#
Fast model creating object masks for real-time image segmentation needs.
Details
Model name:
yolov8s-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 22.79 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8s-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8s-world-torch#
Lightweight detector finding objects based on text descriptions for edge applications.
Details
Model name:
yolov8s-world-torch
Model source: https://docs.ultralytics.com/models/yolo-world/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 25.91 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8s-world-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yolov8s-world-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yolov8x-coco-torch#
High-accuracy detector for critical applications where precision matters most.
Details
Model name:
yolov8x-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 130.53 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8x-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8x-obb-dotav1-torch#
High-precision detector for rotated objects in aerial and satellite imagery analysis.
Details
Model name:
yolov8x-obb-dotav1-torch
Model source: https://docs.ultralytics.com/tasks/obb/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 133.07 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, polylines, obb, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8x-obb-dotav1-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8x-oiv7-torch#
Accurate general detector recognizing over six hundred different object types.
Details
Model name:
yolov8x-oiv7-torch
Model source: https://docs.ultralytics.com/datasets/detect/open-images-v7
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 130.53 MB
Exposes embeddings? no
Tags:
detection, oiv7, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8x-oiv7-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8x-seg-coco-torch#
High-accuracy model generating detailed object outlines for demanding professional applications.
Details
Model name:
yolov8x-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolov8/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 137.40 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8x-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov8x-world-torch#
Open-vocabulary detector with high accuracy for text-based object finding.
Details
Model name:
yolov8x-world-torch
Model source: https://docs.ultralytics.com/models/yolo-world/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 141.11 MB
Exposes embeddings? no
Tags:
detection, torch, yolo, zero-shot, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov8x-world-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
17
18#
19# Make zero-shot predictions with custom classes
20#
21
22model = foz.load_zoo_model(
23 "yolov8x-world-torch",
24 classes=["person", "dog", "cat", "bird", "car", "tree", "chair"],
25)
26
27dataset.apply_model(model, label_field="predictions")
28session.refresh()
yolov9c-coco-torch#
Detector enhanced with transformer technology for improved object finding capabilities.
Details
Model name:
yolov9c-coco-torch
Model source: https://docs.ultralytics.com/models/yolov9/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 49.40 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov9c-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov9c-seg-coco-torch#
Compact model producing both masks and boxes with transformer-enhanced accuracy.
Details
Model name:
yolov9c-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolov9/#__tabbed_1_2
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 53.86 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.42
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov9c-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov9e-coco-torch#
Advanced detector with transformer backbone delivering superior accuracy for complex scenes.
Details
Model name:
yolov9e-coco-torch
Model source: https://docs.ultralytics.com/models/yolov9/
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 112.09 MB
Exposes embeddings? no
Tags:
detection, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.0
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov9e-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
yolov9e-seg-coco-torch#
Advanced model creating precise object outlines using enhanced transformer architecture.
Details
Model name:
yolov9e-seg-coco-torch
Model source: https://docs.ultralytics.com/models/yolov9/#__tabbed_1_2
Model author: Glenn Jocher, et al.
Model license: AGPL-3.0
Model size: 116.55 MB
Exposes embeddings? no
Tags:
instances, coco, torch, yolo, official
Requirements
Packages:
torch>=1.7.0, torchvision>=0.8.1, ultralytics>=8.1.42
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolov9e-seg-coco-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
zero-shot-classification-transformer-torch#
Finds any object you name in images without requiring training on those specific items.
Details
Model name:
zero-shot-classification-transformer-torch
Model source: https://huggingface.co/docs/transformers/tasks/zero_shot_image_classification
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? yes
Tags:
classification, logits, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]
13
14model = foz.load_zoo_model(
15 "zero-shot-classification-transformer-torch",
16 classes=classes,
17)
18
19dataset.apply_model(model, label_field="predictions")
20
21session = fo.launch_app(dataset)
22
23# some models make require additional arguments
24# check the Hugging Face docs to see if any are needed
25
26# for example, AltCLIP requires `padding=True` in its processor
27model = foz.load_zoo_model(
28 "zero-shot-classification-transformer-torch",
29 classes=classes,
30 name_or_path="BAAI/AltCLIP",
31 transformers_processor_kwargs={
32 "padding": True,
33 }
34)
35
36dataset.apply_model(model, label_field="predictions")
37
38session = fo.launch_app(dataset)
zero-shot-detection-transformer-torch#
Hugging Face Transformers model for zero-shot object detection.
Details
Model name:
zero-shot-detection-transformer-torch
Model source: https://huggingface.co/docs/transformers/tasks/zero_shot_object_detection
Model author: Thomas Wolf, et al.
Model license: Apache 2.0
Exposes embeddings? yes
Tags:
detection, logits, embeddings, torch, transformers, zero-shot, official
Requirements
Packages:
torch, torchvision, transformers
CPU support
yes
GPU support
yes
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]
13
14model = foz.load_zoo_model(
15 "zero-shot-detection-transformer-torch",
16 classes=classes,
17)
18
19dataset.apply_model(model, label_field="predictions")
20
21session = fo.launch_app(dataset)
TensorFlow models#
centernet-hg104-1024-coco-tf2#
CenterNet model from Objects as Points with the Hourglass-104 backbone trained on COCO resized to 1024x1024.
Details
Model name:
centernet-hg104-1024-coco-tf2
Model source: tensorflow/models
Model author: Xingyi Zhou, et al.
Model license: Apache 2.0
Model size: 1.33 GB
Exposes embeddings? no
Tags:
detection, coco, tf2, centernet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("centernet-hg104-1024-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
centernet-hg104-512-coco-tf2#
CenterNet model from Objects as Points with the Hourglass-104 backbone trained on COCO resized to 512x512.
Details
Model name:
centernet-hg104-512-coco-tf2
Model source: tensorflow/models
Model author: Xingyi Zhou, et al.
Model license: Apache 2.0
Model size: 1.49 GB
Exposes embeddings? no
Tags:
detection, coco, tf2, centernet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("centernet-hg104-512-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
centernet-mobilenet-v2-fpn-512-coco-tf2#
CenterNet model from Objects as Points with the MobileNetV2 backbone trained on COCO resized to 512x512.
Details
Model name:
centernet-mobilenet-v2-fpn-512-coco-tf2
Model source: tensorflow/models
Model author: Xingyi Zhou, et al.
Model license: Apache 2.0
Model size: 41.98 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, centernet, mobilenet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("centernet-mobilenet-v2-fpn-512-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
centernet-resnet101-v1-fpn-512-coco-tf2#
CenterNet model from Objects as Points with the ResNet-101v1 backbone + FPN trained on COCO resized to 512x512.
Details
Model name:
centernet-resnet101-v1-fpn-512-coco-tf2
Model source: tensorflow/models
Model author: Xingyi Zhou, et al.
Model license: Apache 2.0
Model size: 329.96 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, centernet, resnet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("centernet-resnet101-v1-fpn-512-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
centernet-resnet50-v1-fpn-512-coco-tf2#
CenterNet model from Objects as Points with the ResNet-50-v1 backbone + FPN trained on COCO resized to 512x512.
Details
Model name:
centernet-resnet50-v1-fpn-512-coco-tf2
Model source: tensorflow/models
Model author: Xingyi Zhou, et al.
Model license: Apache 2.0
Model size: 194.61 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, centernet, resnet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("centernet-resnet50-v1-fpn-512-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
centernet-resnet50-v2-512-coco-tf2#
CenterNet model from Objects as Points with the ResNet-50v2 backbone trained on COCO resized to 512x512.
Details
Model name:
centernet-resnet50-v2-512-coco-tf2
Model source: tensorflow/models
Model author: Xingyi Zhou, et al.
Model license: Apache 2.0
Model size: 226.95 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, centernet, resnet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("centernet-resnet50-v2-512-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
deeplabv3-cityscapes-tf#
DeepLabv3+ semantic segmentation model from Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation with Xception backbone trained on the Cityscapes dataset.
Details
Model name:
deeplabv3-cityscapes-tf
Model source: tensorflow/models
Model author: Liang-Chieh Chen, et al.
Model license: Apache 2.0
Model size: 158.04 MB
Exposes embeddings? no
Tags:
segmentation, cityscapes, tf, deeplabv3, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("deeplabv3-cityscapes-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
deeplabv3-mnv2-cityscapes-tf#
DeepLabv3+ semantic segmentation model from Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation with MobileNetV2 backbone trained on the Cityscapes dataset.
Details
Model name:
deeplabv3-mnv2-cityscapes-tf
Model source: tensorflow/models
Model author: Liang-Chieh Chen, et al.
Model license: Apache 2.0
Model size: 8.37 MB
Exposes embeddings? no
Tags:
segmentation, cityscapes, tf, deeplabv3, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("deeplabv3-mnv2-cityscapes-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d0-512-coco-tf2#
EfficientDet-D0 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 512x512.
Details
Model name:
efficientdet-d0-512-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 29.31 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d0-512-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d0-coco-tf1#
EfficientDet-D0 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO.
Details
Model name:
efficientdet-d0-coco-tf1
Model source: voxel51/automl
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 38.20 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, efficientdet, legacy
Requirements
CPU support
yes
Packages:
tensorflow>=1.14,<2
GPU support
yes
Packages:
tensorflow-gpu>=1.14,<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d0-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d1-640-coco-tf2#
EfficientDet-D1 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 640x640.
Details
Model name:
efficientdet-d1-640-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 49.44 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d1-640-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d1-coco-tf1#
EfficientDet-D1 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO.
Details
Model name:
efficientdet-d1-coco-tf1
Model source: voxel51/automl
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 61.64 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, efficientdet, legacy
Requirements
CPU support
yes
Packages:
tensorflow>=1.14,<2
GPU support
yes
Packages:
tensorflow-gpu>=1.14,<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d1-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d2-768-coco-tf2#
EfficientDet-D2 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 768x768.
Details
Model name:
efficientdet-d2-768-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 60.01 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d2-768-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d2-coco-tf1#
EfficientDet-D2 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO.
Details
Model name:
efficientdet-d2-coco-tf1
Model source: voxel51/automl
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 74.00 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, efficientdet, legacy
Requirements
CPU support
yes
Packages:
tensorflow>=1.14,<2
GPU support
yes
Packages:
tensorflow-gpu>=1.14,<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d2-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d3-896-coco-tf2#
EfficientDet-D3 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 896x896.
Details
Model name:
efficientdet-d3-896-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 88.56 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d3-896-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d3-coco-tf1#
EfficientDet-D3 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO.
Details
Model name:
efficientdet-d3-coco-tf1
Model source: voxel51/automl
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 106.44 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, efficientdet, legacy
Requirements
CPU support
yes
Packages:
tensorflow>=1.14,<2
GPU support
yes
Packages:
tensorflow-gpu>=1.14,<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d3-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d4-1024-coco-tf2#
EfficientDet-D4 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 1024x1024.
Details
Model name:
efficientdet-d4-1024-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 151.15 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d4-1024-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d4-coco-tf1#
EfficientDet-D4 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO.
Details
Model name:
efficientdet-d4-coco-tf1
Model source: voxel51/automl
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 175.33 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, efficientdet, legacy
Requirements
CPU support
yes
Packages:
tensorflow>=1.14,<2
GPU support
yes
Packages:
tensorflow-gpu>=1.14,<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d4-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d5-1280-coco-tf2#
EfficientDet-D5 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 1280x1280.
Details
Model name:
efficientdet-d5-1280-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 244.41 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d5-1280-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d5-coco-tf1#
EfficientDet-D5 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO.
Details
Model name:
efficientdet-d5-coco-tf1
Model source: voxel51/automl
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 275.81 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, efficientdet, legacy
Requirements
CPU support
yes
Packages:
tensorflow>=1.14,<2
GPU support
yes
Packages:
tensorflow-gpu>=1.14,<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d5-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d6-1280-coco-tf2#
EfficientDet-D6 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 1280x1280.
Details
Model name:
efficientdet-d6-1280-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 375.63 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d6-1280-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d6-coco-tf1#
EfficientDet-D6 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO.
Details
Model name:
efficientdet-d6-coco-tf1
Model source: voxel51/automl
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 416.43 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, efficientdet, legacy
Requirements
CPU support
yes
Packages:
tensorflow>=1.14,<2
GPU support
yes
Packages:
tensorflow-gpu>=1.14,<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d6-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
efficientdet-d7-1536-coco-tf2#
EfficientDet-D7 model from EfficientDet: Scalable and Efficient Object Detection trained on COCO resized to 1536x1536.
Details
Model name:
efficientdet-d7-1536-coco-tf2
Model source: tensorflow/models
Model author: Mingxing Tan, et al.
Model license: Apache 2.0
Model size: 376.20 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, efficientdet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("efficientdet-d7-1536-coco-tf2")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-inception-resnet-atrous-v2-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks atrous version with Inception backbone trained on COCO.
Details
Model name:
faster-rcnn-inception-resnet-atrous-v2-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 234.46 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, inception, resnet
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-inception-resnet-atrous-v2-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-inception-resnet-atrous-v2-lowproposals-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks atrous version with low-proposals and Inception backbone trained on COCO.
Details
Model name:
faster-rcnn-inception-resnet-atrous-v2-lowproposals-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 234.46 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, inception, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-inception-resnet-atrous-v2-lowproposals-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-inception-v2-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks with Inception v2 backbone trained on COCO.
Details
Model name:
faster-rcnn-inception-v2-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 52.97 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, inception
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-inception-v2-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-nas-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks with NAS-net backbone trained on COCO.
Details
Model name:
faster-rcnn-nas-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 404.95 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-nas-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-nas-lowproposals-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks with low-proposals and NAS-net backbone trained on COCO.
Details
Model name:
faster-rcnn-nas-lowproposals-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 404.88 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-nas-lowproposals-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-resnet101-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks with ResNet-101 backbone trained on COCO.
Details
Model name:
faster-rcnn-resnet101-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 186.41 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-resnet101-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-resnet101-lowproposals-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks with low-proposals and ResNet-101 backbone trained on COCO.
Details
Model name:
faster-rcnn-resnet101-lowproposals-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 186.41 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-resnet101-lowproposals-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-resnet50-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks with ResNet-50 backbone trained on COCO.
Details
Model name:
faster-rcnn-resnet50-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 113.57 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, resnet
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-resnet50-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
faster-rcnn-resnet50-lowproposals-coco-tf#
Faster R-CNN model from Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks with low-proposals and ResNet-50 backbone trained on COCO.
Details
Model name:
faster-rcnn-resnet50-lowproposals-coco-tf
Model source: tensorflow/models
Model author: Shaoqing Ren, et al.
Model license: Apache 2.0
Model size: 113.57 MB
Exposes embeddings? no
Tags:
detection, coco, tf, faster-rcnn, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("faster-rcnn-resnet50-lowproposals-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
inception-resnet-v2-imagenet-tf1#
Inception v2 model from Rethinking the Inception Architecture for Computer Vision trained on ImageNet.
Details
Model name:
inception-resnet-v2-imagenet-tf1
Model source: tensorflow/models
Model author: Christian Szegedy, et al.
Model license: Apache 2.0
Model size: 213.81 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, tf1, inception, resnet
Requirements
CPU support
yes
Packages:
tensorflow<2
GPU support
yes
Packages:
tensorflow-gpu<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("inception-resnet-v2-imagenet-tf1")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
inception-v4-imagenet-tf1#
Inception v4 model from Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning trained on ImageNet.
Details
Model name:
inception-v4-imagenet-tf1
Model source: tensorflow/models
Model author: Christian Szegedy, et al.
Model license: Apache 2.0
Model size: 163.31 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, tf1, inception, legacy
Requirements
CPU support
yes
Packages:
tensorflow<2
GPU support
yes
Packages:
tensorflow-gpu<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("inception-v4-imagenet-tf1")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
mask-rcnn-inception-resnet-v2-atrous-coco-tf#
Mask R-CNN model from Mask R-CNN atrous version with Inception backbone trained on COCO.
Details
Model name:
mask-rcnn-inception-resnet-v2-atrous-coco-tf
Model source: tensorflow/models
Model author: Kaiming He, et al.
Model license: Apache 2.0
Model size: 254.51 MB
Exposes embeddings? no
Tags:
instances, coco, tf, mask-rcnn, inception, resnet
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("mask-rcnn-inception-resnet-v2-atrous-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
mask-rcnn-inception-v2-coco-tf#
Mask R-CNN model from Mask R-CNN with Inception backbone trained on COCO.
Details
Model name:
mask-rcnn-inception-v2-coco-tf
Model source: tensorflow/models
Model author: Kaiming He, et al.
Model license: Apache 2.0
Model size: 64.03 MB
Exposes embeddings? no
Tags:
instances, coco, tf, mask-rcnn, inception
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("mask-rcnn-inception-v2-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
mask-rcnn-resnet101-atrous-coco-tf#
Mask R-CNN model from Mask R-CNN atrous version with ResNet-101 backbone trained on COCO.
Details
Model name:
mask-rcnn-resnet101-atrous-coco-tf
Model source: tensorflow/models
Model author: Kaiming He, et al.
Model license: Apache 2.0
Model size: 211.56 MB
Exposes embeddings? no
Tags:
instances, coco, tf, mask-rcnn, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("mask-rcnn-resnet101-atrous-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
mask-rcnn-resnet50-atrous-coco-tf#
Mask R-CNN model from Mask R-CNN atrous version with ResNet-50 backbone trained on COCO.
Details
Model name:
mask-rcnn-resnet50-atrous-coco-tf
Model source: tensorflow/models
Model author: Kaiming He, et al.
Model license: Apache 2.0
Model size: 138.29 MB
Exposes embeddings? no
Tags:
instances, coco, tf, mask-rcnn, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("mask-rcnn-resnet50-atrous-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
mobilenet-v2-imagenet-tf1#
MobileNetV2 model from MobileNetV2: Inverted Residuals and Linear Bottlenecks trained on ImageNet.
Details
Model name:
mobilenet-v2-imagenet-tf1
Model source: tensorflow/models
Model author: Mark Sandler, et al.
Model license: Apache 2.0
Model size: 13.64 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, tf1, mobilenet
Requirements
CPU support
yes
Packages:
tensorflow<2
GPU support
yes
Packages:
tensorflow-gpu<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("mobilenet-v2-imagenet-tf1")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnet-v1-50-imagenet-tf1#
ResNet-50 v1 model from Deep Residual Learning for Image Recognition trained on ImageNet.
Details
Model name:
resnet-v1-50-imagenet-tf1
Model source: tensorflow/models
Model author: Kaiming He, et al.
Model license: Apache 2.0
Model size: 97.84 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, tf1, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow<2
GPU support
yes
Packages:
tensorflow-gpu<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnet-v1-50-imagenet-tf1")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
resnet-v2-50-imagenet-tf1#
ResNet-50 v2 model from Deep Residual Learning for Image Recognition trained on ImageNet.
Details
Model name:
resnet-v2-50-imagenet-tf1
Model source: tensorflow/models
Model author: Kaiming He, et al.
Model license: Apache 2.0
Model size: 97.86 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, tf1, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow<2
GPU support
yes
Packages:
tensorflow-gpu<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("resnet-v2-50-imagenet-tf1")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
rfcn-resnet101-coco-tf#
R-FCN object detection model from R-FCN: Object Detection via Region-based Fully Convolutional Networks with ResNet-101 backbone trained on COCO.
Details
Model name:
rfcn-resnet101-coco-tf
Model source: tensorflow/models
Model author: Jifeng Dai, et al.
Model license: Apache 2.0
Model size: 208.16 MB
Exposes embeddings? no
Tags:
detection, coco, tf, rfcn, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("rfcn-resnet101-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
ssd-inception-v2-coco-tf#
Inception Single Shot Detector model from SSD: Single Shot MultiBox Detector trained on COCO.
Details
Model name:
ssd-inception-v2-coco-tf
Model source: tensorflow/models
Model author: Wei Liu, et al.
Model license: Apache 2.0
Model size: 97.50 MB
Exposes embeddings? no
Tags:
detection, coco, tf, ssd, inception
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("ssd-inception-v2-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
ssd-mobilenet-v1-coco-tf#
Single Shot Detector model from SSD: Single Shot MultiBox Detector with MobileNetV1 backbone trained on COCO.
Details
Model name:
ssd-mobilenet-v1-coco-tf
Model source: tensorflow/models
Model author: Wei Liu, et al.
Model license: Apache 2.0
Model size: 27.83 MB
Exposes embeddings? no
Tags:
detection, coco, tf, ssd, mobilenet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("ssd-mobilenet-v1-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
ssd-mobilenet-v1-fpn-640-coco17#
MobileNetV1 model from MobileNetV2: Inverted Residuals and Linear Bottlenecks resized to 640x640.
Details
Model name:
ssd-mobilenet-v1-fpn-640-coco17
Model source: tensorflow/models
Model author: Mark Sandler, et al.
Model license: Apache 2.0
Model size: 43.91 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, ssd, mobilenet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("ssd-mobilenet-v1-fpn-640-coco17")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
ssd-mobilenet-v1-fpn-coco-tf#
FPN Single Shot Detector model from SSD: Single Shot MultiBox Detector with MobileNetV1 backbone trained on COCO.
Details
Model name:
ssd-mobilenet-v1-fpn-coco-tf
Model source: tensorflow/models
Model author: Wei Liu, et al.
Model license: Apache 2.0
Model size: 48.97 MB
Exposes embeddings? no
Tags:
detection, coco, tf, ssd, mobilenet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("ssd-mobilenet-v1-fpn-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
ssd-mobilenet-v2-320-coco17#
MobileNetV2 model from MobileNetV2: Inverted Residuals and Linear Bottlenecks resized to 320x320.
Details
Model name:
ssd-mobilenet-v2-320-coco17
Model source: tensorflow/models
Model author: Mark Sandler, et al.
Model license: Apache 2.0
Model size: 43.91 MB
Exposes embeddings? no
Tags:
detection, coco, tf2, ssd, mobilenet
Requirements
CPU support
yes
Packages:
tensorflow>=2|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu>=2|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("ssd-mobilenet-v2-320-coco17")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
ssd-resnet50-fpn-coco-tf#
FPN Single Shot Detector model from SSD: Single Shot MultiBox Detector with ResNet-50 backbone trained on COCO.
Details
Model name:
ssd-resnet50-fpn-coco-tf
Model source: tensorflow/models
Model author: Wei Liu, et al.
Model license: Apache 2.0
Model size: 128.07 MB
Exposes embeddings? no
Tags:
detection, coco, tf, ssd, resnet, legacy
Requirements
CPU support
yes
Packages:
tensorflow|tensorflow-macos
GPU support
yes
Packages:
tensorflow-gpu|tensorflow>=2|tensorflow-macos
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("ssd-resnet50-fpn-coco-tf")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)
vgg16-imagenet-tf1#
VGG-16 model from Very Deep Convolutional Networks for Large-Scale Image Recognition trained on ImageNet.
Details
Model name:
vgg16-imagenet-tf1
Model source: https://gist.github.com/ksimonyan/211839e770f7b538e2d8#file-readme-md
Model author: Karen Simonyan, et al.
Model license: CC-BY-4.0
Model size: 527.80 MB
Exposes embeddings? yes
Tags:
classification, embeddings, logits, imagenet, tf1, vgg, legacy
Requirements
CPU support
yes
Packages:
tensorflow<2
GPU support
yes
Packages:
tensorflow-gpu<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "imagenet-sample",
6 dataset_name=fo.get_default_dataset_name(),
7 max_samples=50,
8 shuffle=True,
9)
10
11model = foz.load_zoo_model("vgg16-imagenet-tf1")
12
13dataset.apply_model(model, label_field="predictions")
14
15session = fo.launch_app(dataset)
yolo-v2-coco-tf1#
YOLOv2 model from YOLO9000: Better, Faster, Stronger trained on COCO.
Details
Model name:
yolo-v2-coco-tf1
Model source: thtrieu/darkflow
Model author: Joseph Redmon, et al.
Model license: GPL-3.0
Model size: 194.49 MB
Exposes embeddings? no
Tags:
detection, coco, tf1, yolo, legacy
Requirements
CPU support
yes
Packages:
tensorflow<2
GPU support
yes
Packages:
tensorflow-gpu<2
Example usage
1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset(
5 "coco-2017",
6 split="validation",
7 dataset_name=fo.get_default_dataset_name(),
8 max_samples=50,
9 shuffle=True,
10)
11
12model = foz.load_zoo_model("yolo-v2-coco-tf1")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)