When I run the following command: ego4d --output_directory="./train_vids/" --datasets full_scale --video_uid_file ./train.txt --version v1
I get the following error:
RuntimeError: The following requested video UIDs could not be found in the manifest for version: 'v1' and dataset(s): '{'full_scale'}':
{'4a987e8e-6ca2-4942-9e07-5b628ef71fdc', '4513f621-033c-4db7-983e-480df9bf6693', '755a28a5-d0c1-4421-9eb5-36ecb33b818c', '0e0b1f0d-adc1-4cb1-9ceb-d5653c6166f4', '7b3c676b-d54b-4e0e-8749-1e1d1ea7b8c6'}
The videos in the video_uid_file (train.txt) come straight from the locomotion benchmark dataset. Could these videos please be uploaded?
Also, we noticed some inconsistencies between the practices reported in the paper and the data provided in the locomotion benchmark. The paper reports "Each video clip must be longer than 10 seconds for past trajectory observation and future prediction." but the video clips in the locomotion data often do not span more than 1 or 2 seconds.
Hi,
When I run the following command: ego4d --output_directory="./train_vids/" --datasets full_scale --video_uid_file ./train.txt --version v1 I get the following error: RuntimeError: The following requested video UIDs could not be found in the manifest for version: 'v1' and dataset(s): '{'full_scale'}': {'4a987e8e-6ca2-4942-9e07-5b628ef71fdc', '4513f621-033c-4db7-983e-480df9bf6693', '755a28a5-d0c1-4421-9eb5-36ecb33b818c', '0e0b1f0d-adc1-4cb1-9ceb-d5653c6166f4', '7b3c676b-d54b-4e0e-8749-1e1d1ea7b8c6'} The videos in the video_uid_file (train.txt) come straight from the locomotion benchmark dataset. Could these videos please be uploaded?
Also, we noticed some inconsistencies between the practices reported in the paper and the data provided in the locomotion benchmark. The paper reports "Each video clip must be longer than 10 seconds for past trajectory observation and future prediction." but the video clips in the locomotion data often do not span more than 1 or 2 seconds.