map-anything-apache-torch#

Universal feed-forward metric 3D reconstruction from images producing depth maps, camera poses, intrinsics, and point clouds.

Details

  • Model name: map-anything-apache-torch

  • Model source: https://huggingface.co/facebook/map-anything-apache

  • Model author: Nikhil Keetha, et al.

  • Model license: Apache 2.0

  • Model size: 3.94 GB

  • Exposes embeddings? no

  • Tags: depth, 3d, pointcloud, reconstruction, torch

Requirements

  • Packages: torch, torchvision, mapanything @ git+https://github.com/facebookresearch/map-anything.git@v1.1

  • 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("map-anything-apache-torch")
13
14dataset.apply_model(model, label_field="predictions")
15
16session = fo.launch_app(dataset)