MOSE: A New Dataset for Video Object Segmentation in Complex Scenes.

Henghui Ding, Chang Liu, Shuting He, Xudong Jiang, Philip H. S. Torr, Song Bai

Browse the full ICCV paper archive.