google-deepmind / google-deepmind/romo
Eval results
Open
- Dominant language
- Python
- Stars
- 40
- Forks
- 4
- PR merge metrics
- No merged PRs in 30d
Description
Hello, I use the provided script to extract motion mask from DAVIS 2016 val dataset and do evaluation on it. In the paper it is 77.3 but I only have
65.9437. When examining the results, I find background are also included in many scenes and also some moving objects are not segmented.
https://github.com/user-attachments/assets/96d1ef57-c0ae-4958-94e8-abcfd78a3d98
https://github.com/user-attachments/assets/89c65532-162e-4047-8cb7-493768c13b76
https://github.com/user-attachments/assets/238755cf-7d13-4186-b07f-04b50a423951
Can you explain why this happen? Also can you provide the your segmentations of the above scenes for my reference? Thanks in advance!
Contributor guide
Assessment
This issue has not been assessed yet.