Scenario
← All Models
Image

SAM 3.1 Image

Use it ↗

Segments any image into isolated object masks. Accepts a text prompt or up to 10 bounding boxes to guide detection. Outputs one PNG mask per object, up to 90+ in complex scenes.

Segment images with Meta SAM 3.1: text and/or box prompting, or interactive tracking (points, box, mask).

More models from Meta