Grounded SAM

Grounded SAM is a demo tool that finds and cuts out objects in images from text descriptions for developers and researchers.

What it is good for

  • Finding and outlining objects in images by name
  • Creating new image edits by replacing parts of a picture
  • Generating labeled training data for image projects
  • Editing images by speaking instructions

At a glance

Who it is for
Setup
Guided command-line setup

Level 4 of 5. You copy and paste commands into a terminal by following the instructions. Comfortable territory for vibe coders.

Cost
Free and open source
Type
Agent app
Tasks
Works with
Grounding DINO, Segment Anything, Stable Diffusion, BLIP, Whisper, ChatGPT

Momentum

Rank this week
Not ranked yet
Best rank
None yet
Hugging Face likes
99

GitHub numbers as of Nov 2, 2026. Updated weekly.

Questions

Is Grounded SAM free?
Yes. It is free and open source.
Do I need to know how to code to use Grounded SAM?
You do not need to write code, but setup means copying and pasting commands into a terminal by following its instructions.
What does Grounded SAM work with?
According to its own documentation, it works with Grounding DINO, Segment Anything, Stable Diffusion, BLIP, Whisper, and ChatGPT.
What is Grounded SAM good for?
Finding and outlining objects in images by name, creating new image edits by replacing parts of a picture, generating labeled training data for image projects, and editing images by speaking instructions.

Similar agents to Grounded SAM

  • BeatDesign

    BeatDesign is a local-first canvas and video editor that lets AI image and video tools and MCP agents build media projects for creators.

  • claude-remotion-skill

    Claude Remotion Skill is a skill that teaches Claude to create and edit motion graphics videos from a single prompt for people who want polished short videos without video software.

  • Oh Story DSH

    Oh Story DSH is a plugin that adds writing workspaces for novels, short dramas, games and video recaps for writers.

  • Open Design

    Open Design is a local-first, open source design workspace where teams and AI coding agents build prototypes, slides, images, and videos together.

  • OpenPencil

    OpenPencil is a vector design tool that turns text prompts into UI on a canvas, with AI agent teams working in parallel for designers and developers.

  • design-extract

    design-extract is a command-line tool that reads a website's design system and exports tokens and code for designers and developers building matching interfaces.

Last updated Nov 2, 2026. Tracked since Oct 28, 2026. Parts of this summary required judgment because the source material was brief. The summary was written by an AI model from the project's own description and checked against our copy rules. Spotted an error? Tell us.