Hello! According to the official introduction of the dataset, there are 6849 clips in total. However, the number of entries in list/hmdb51/train_rgb_split_[123].txt is 3570, and the number of entries in list/hmdb51/val_rgb_split_[123].txt is 1530. It seems that not all clips are being used. Could you please explain the reasoning behind the removal of certain data?
Hello! According to the official introduction of the dataset, there are 6849 clips in total. However, the number of entries in
list/hmdb51/train_rgb_split_[123].txtis 3570, and the number of entries inlist/hmdb51/val_rgb_split_[123].txtis 1530. It seems that not all clips are being used. Could you please explain the reasoning behind the removal of certain data?