Overview
SAM 3 extends SAM 2 from clicking on one object to “promptable concept segmentation”: given a phrase such as “yellow school bus” or an exemplar box, it finds and masks all matching objects and can track them through video. It still accepts points, boxes and masks as prompts.
Highlights
- Text and exemplar prompts for open-vocabulary segmentation
- Points, boxes and masks as visual prompts
- Image and video segmentation with tracking
- Supported in Hugging Face Transformers
Typical uses
Data labeling, video editing, visual search, robotics perception and analytics.
Things to consider
- Weights are gated behind Meta’s SAM License; review its terms.
- Short noun phrases work best as text prompts.
Specifications are taken from the model card on Hugging Face and the provider’s announcement. Benchmark results depend on evaluation settings; see how AI benchmarks work.