classifies one frame at a time?

[xdsv]

Hi,

does miles-deep classify each frame at a time individually or does it use the temporal information (treats it as a sequence of images)?

[ryanjay0]

Miles Deep only looks at individual frames for classification. However, when making cut blocks it does some simple smoothing to avoid small gaps.

Karpathy et al. give some ways to extend it to to video but the results aren't that much better: https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42455.pdf

Some recent work gives more insight about video classification: https://arxiv.org/abs/1708.03805 and the results are encouraging but i still doubt it would do much better than a single frame model. Let me know if you try anything like this.