Confused about video processing

[chelsea-6]

I'm currently working on video preprocessing for the thumos14 dataset, taking the video_validation_0000053 as an example, as far as I know it has a total of 5916 frames, and going by the method you described to divide the video, with a step of 64 code snippets, it doesn't seem to be able to divide the subvideos from 0-17, and I see that you were able to end up dividing them from 0-17, are there Is there another way to handle the details? I'm very confused, could you please help me with this?