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

MOSE is a new dataset for video object segmentation in complex scenes. It contains 2149 videos, out of which only 1507 come with available ground-truth masks.

Data and Resources

Cite this as

Henghui Ding, Chang Liu, Shuting He, Xudong Jiang, Philip HS Torr, Song Bai (2024). Dataset: MOSE: A New Dataset for Video Object Segmentation in Complex Scenes. https://doi.org/10.57702/8imxc3e9

DOI retrieved: December 2, 2024

Additional Info

Field Value
Created December 2, 2024
Last update December 2, 2024
Defined In https://doi.org/10.48550/arXiv.2311.04414
Author Henghui Ding
More Authors
Chang Liu
Shuting He
Xudong Jiang
Philip HS Torr
Song Bai
Homepage https://github.com/henghui-Ding/MOSE