Back to Apps

MCP Vision
by groundlight
Expose HuggingFace computer vision models as MCP tools for zero-shot object detection and zooming.
0 stars
Works in:claude
Exposes:Tools
What it does
MCP Vision integrates state-of-the-art HuggingFace computer vision models into AI agents via the Model Context Protocol. It allows LLMs to perform zero-shot object detection and interactive image cropping (zooming) to analyze visual data with high precision.
Tools
- locate_objects: Detects and locates objects in an image using zero-shot pipelines (default: owlvit-large-patch14).
- zoom_to_object: Crops the image to the bounding box of a specific label for closer analysis.
Installation
Build the local docker image:
git clone git@github.com:groundlight/mcp-vision.git && cd mcp-vision && make build-docker
Supported hosts
- claude
Quick install
make build-dockerInformation
- Pricing
- free
- Published
- 8/4/2026
- stars
- 0
Categories
Choose your AI client and follow the steps below.
Claude Desktop
{
"mcpServers": {
"mcp-vision": {
"command": "docker",
"args": ["run", "-i", "--rm", "mcp-vision"]
}
}
}




