Any recommendations for video (motion-based) machine-learning?

I have been looking into machine learning tools for a while, esp. pyTorch and scikitlearn and want to use them for object identification. All the resources I have found so far, however, deal with image recognition rather than identification based on position, trajectory, etc. For example, even though a bird and a plane may appear similar at a distance from the camera, the flight trajectories across the video can be used to manually distinguish one from the other. Can anyone recommend a resource for motion-based machine learning tools for object identification?


so you want to… classify the trajectory? I guess you can do that. treat the coordinate tuples like text data, or like audio data (time series data).

Thank you. I will take a look at those options.