Closed priydarshi1o1 closed 4 years ago
You can use our algorithm to get objects masks for the first frame. Then you can try to use some sort of tracking with segmentation algorithms (e.g. MOTS: Multi-Object Tracking and Segmentation).
Thanks for your reply, although MOTS authors have provided scripts for traning for semi-interactive annotation they haven't provided anything.
For sequence dataset(video) can we use this interactive segmentation for using previously segmented objects in previous frames.