Was the video classification example trained from scratch? Or were the weights from I3d module used?
Was the video classification example trained from scratch? Or were the weights from I3d module used?