Open NNNNAI opened 2 years ago
Hi, I used following command to download data related to EM task which is in size of 2.75 TB.
python -m ego4d.cli.cli --output_directory="/mnt/g/Episodic_Memory/Ego4D" --datasets full_scale annotations vq2d_models 3d 3d_scans 3d_scan_keypoints --benchmarks EM
If you exclude options related to 3d and models, it may reduce the size. At the time of my download, I used only the options available in the page https://github.com/facebookresearch/Ego4d/blob/main/ego4d/cli/README.md
there may be undocumented options to select subsets like VQ2D, VQ3D, NLQ and MQ in the EM task. if not changes to scripts required to filter the data download.
@NNNNAI: There is already similar issue created here https://github.com/EGO4D/episodic-memory/issues/3 Read the comment https://github.com/EGO4D/episodic-memory/issues/3#issuecomment-1080165970 for the work around.
Hi, thanks for sharing your work. In the download sricpt, the choice of --benchmark is 'EM', 'FHO' and 'AV'. How can I download the the NLQ subset without downloading all the video of 'EM'? Thanks for your time, have a nice day.