I tried to use your implementation, but I couldn't find out where to adapt.
- Could you maybe provide the necessary steps to reproduce the results of the paper?
- Do I have to first train the model or do you provide the trained model?
- How can I test the method on a custom video?
Thanks!