This is a python script for AVA Actions Dataset Videos downloading.
- The video name lists of train/val data and test data are saved in "train_val_video_names.txt" and "test_video_names.txt" respectively.
- run "download_videos.py" to download the videos.
- To download this dataset, about 212 GB of memory is required
- Several Videos may be failed for some reason, run "download_videos.py" to re-download the failed files.
The annotation of the dataset is available at https://s3.amazonaws.com/ava-dataset/annotations/ava_v2.2.zip.