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)