facebookresearch / facebookresearch/sam2
Some blocks in SAM2 have particularly low confidence, resulting in some blocks being missed. Is there any solution available?
Open
- Dominant language
- Jupyter Notebook
- Stars
- 19.9k
- Forks
- 2.5k
- PR merge metrics
- No merged PRs in 30d
Description
Our experiment used grid point sampling, but due to the confidence threshold and IOU threshold, some blocks were missed, which are what we need. Is there any way to solve this problem?
Contributor guide
Assessment
This issue has not been assessed yet.