Closed LjIA26 closed 1 year ago
Hi! Thanks for your interest - this model would work on images where the main differences between images are the parts you want to track. Notably, videos with stationary backgrounds have this property, but if you had an image collection where this property holds, it would also work. Let me know if you have any other questions!
If there's no other questions, I will close this issue for now.
Hello! Great work! I was wondering if there was a way to use your model for images instead of videos for keypoint discovery? Thank you.