History

README.md

dgp/cli.py is the main CLI entrypoint for handling DGP datasets.

DGP CLI subcommandsvisualize-scenes andvisualize-scene can be used tovisualize DGP-compliant data.

• To visualize a split of aDGP SceneDataset,runpython dgp/cli.py visualize-scenes:

  Show the help message via:

  dgp$ python dgp/cli.py visualize-scenes --help

  Example command to visualize the images fromCAMERA_01, CAMERA_05, CAMERA_06and point cloud fromLIDAR along with ground_truth annotationsbounding_box_2d, bounding_box_3d fromtrain split of the toy datasettests/data/dgp/test_scene/scene_dataset_v1.0.json, and store the resultingvideos in--dst-dir vis. One can find the resulting 3D visualization videosinvis/3d and 2D visualization videos invis/2d.

  dgp$ python dgp/cli.py visualize-scenes --scene-dataset-json tests/data/dgp/test_scene/scene_dataset_v1.0.json --split train --dst-dir vis -l LIDAR -c CAMERA_01 -c CAMERA_05 -c CAMERA_06 -a bounding_box_2d -a bounding_box_3d

  

Add flagrender-pointcloud to render projected pointcloud onto images:

dgp$ python dgp/cli.py visualize-scenes --scene-dataset-json tests/data/dgp/test_scene/scene_dataset_v1.0.json --split train --dst-dir vis -l LIDAR -c CAMERA_01 -c CAMERA_05 -c CAMERA_06 -a bounding_box_2d -a bounding_box_3d --render-pointcloud

• To visualize a singleDGP Scene, runpython dgp/cli.py visualize-scene:

  Show the help message via:

  dgp$ python dgp/cli.py visualize-scene --help

  Example command to visualize the images fromCAMERA_01, CAMERA_05, CAMERA_06and point cloud fromLIDAR along with ground_truth annotationsbounding_box_2d, bounding_box_3d from the toy Scenetests/data/dgp/test_scene/scene_01/scene_a8dc5ed1da0923563f85ea129f0e0a83e7fe1867.json,and store the resulting videos in--dst-dir vis. One can find the resulting3D visualization video invis/3d and 2D visualization video invis/2d.

  dgp$ python dgp/cli.py visualize-scene --scene-json tests/data/dgp/test_scene/scene_01/scene_a8dc5ed1da0923563f85ea129f0e0a83e7fe1867.json --dst-dir vis -l LIDAR -c CAMERA_01 -c CAMERA_05 -c CAMERA_06 -a bounding_box_2d -a bounding_box_3d

DGP CLI provides information about a dataset, including the remote location (S3url) of the dataset, its raw dataset url, the set of available annotation typescontained in the dataset, etc. For more information, see relevant metadatastored with a dataset artifact inDatasetMetadata andDatasetArtifacts.

dgp$ python dgp/cli.py info --scene-dataset-json<scene-dataset-json>

DGP CLI provides a simplified mechanism for validating newly created datasets,ensuring that the dataset schema is maintained and valid. This is done via:

dgp$ python dgp/cli.py validate --scene-dataset-json<scene-dataset-json>