Language Segment-Anything
lastmod 2026-11-10created 2024-05-032 edits
Language Segment-Anything is an open-source project that segments objects in an image from a text prompt. It has a two-stage structure: Grounding DINO finds bounding boxes of objects matching the text description, and those boxes are fed into SAM (Segment Anything Model) to obtain precise masks. It lets SAM, which otherwise requires clicking points or drawing boxes, be used with natural language input.
SAM with text prompt