I have recently come across the Kaggle Deepfake Detection Challenge and I was wondering if we can use fastai to do video classification at all?
I am wondering this as well. As far as I can tell fastai does not have native support for lstm’s on image data only for text data. This would make it difficult to take into account the temporal information from the video but you can still a regular cnn on the still frames. I would love to be wrong about all this though…