facebookresearch / facebookresearch/segment-anything
Automatic mask generator for area of interest
Nobody has claimed this yet.
- Dominant language
- Jupyter Notebook
- Stars
- 54.9k
- Forks
- 6.4k
- PR merge metrics
- No merged PRs in 30d
Description
Hi. I'm using AutomaticMaskGenerator for segment ct images. The model mostly failed to do that, due to the black background of the image, for example:
I tried to use the points prompting, I've marked the black background with 0 labels, and the foreground of the ct image with label 1. But this method only returns 3 masks, and I need it to return every mask that it can.
Are there any ways to focus the AutomaticMaskGenerator on the foreground of the image to segment more masks?
another example of a liver ultrasound image:
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with AutomaticMaskGenerator and the points-prompting behavior described in the issue. Reproduce the CT and liver-ultrasound examples, then determine whether focusing on the foreground can produce all available masks; done means a documented or implemented approach that addresses the requested foreground and mask count.
Written by the indexing model from the issue text.
Assessment
- Domain
- computer-vision, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100