I want to train a Computer Vision model using videos. I have a 39-hour long video on YouTube that I’d like to use to train a CV model which will detect potholes and other defects on the road. A user on this forum suggested the MaskPropagation method from this paper
I’ve never implemented a research paper before. What is the best way to implement this so I can apply it to my model?