Following #93, we have to implement functionality to support this frames.csv. This will essentially tell the frame sampler that these are the frames that are available instead of using all the frames in the video. This means that we will ONLY sample from the keyframes if this csv file is EXPLICITLY provided. It does not load it implicitly.
Format for the csv:
video_id.mp4 1 22 34
video_id2.mp4 12 24 35
This should be stored in a CSV file, either at the root level of the dataset or provided as a full path to the frame.csv file.
Following #93, we have to implement functionality to support this frames.csv. This will essentially tell the frame sampler that these are the frames that are available instead of using all the frames in the video. This means that we will ONLY sample from the keyframes if this csv file is EXPLICITLY provided. It does not load it implicitly.
Format for the csv:
This should be stored in a CSV file, either at the root level of the dataset or provided as a full path to the frame.csv file.