Closed svenvanderburg closed 2 years ago
Not all videos have the same length, some are shorter. For engineered feature experiments this is not a problem, because we just calculate statistics over the short video. For deep learning this is a problem though.
There is about 10 videos for lying data that have less than 501 frames. We choose to clip and pad the data for deep learning to 501 frames.
Implemented with https://github.com/RehabAUmc/modys-video/pull/11
Not all videos have the same length, some are shorter. For engineered feature experiments this is not a problem, because we just calculate statistics over the short video. For deep learning this is a problem though.