facebookresearch / facebookresearch/segment-anything
How to choose the suitable mask?
Open
- Dominant language
- Jupyter Notebook
- Stars
- 54.9k
- Forks
- 6.4k
- PR merge metrics
- No merged PRs in 30d
Description
When I use SAM to segment medical images in an end-to-end manner using point prompts, the segmented labels are not easily identified correctly.
When I add the point promptly, SAM cancels out the whole image more, resulting in a large score.
Is there any way to select a suitable mask after adding the point prompt to guide SAM to split the label correctly?
Contributor guide
Assessment
This issue has not been assessed yet.