MCP Server for CVDLT(Computer Vision & Deep Learning Tools)

Name: MCP Server for CVDLT(Computer Vision & Deep Learning Tools)
Author: MRonaldo-gif

@MRonaldo-gif

About MCP Server for CVDLT(Computer Vision & Deep Learning Tools)

The repo is based on Model Context procotol of Python SDK, including DL models in CV, and provide the abilities to the LLM or vLLM model

#mcp-server #computer-vision #deep-learning

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "mcp-server-cvdlt": {
      "command": "uv",
      "args": [
        "sync"
      ]
    }
  }
}

Tools

Detect objects in an image using YOLOv10

Segment objects in an image using YOLOv8

Segment entire image using Ultralytics SAM

Estimate human poses in an image using YOLOv8

Overview

What is MCP Server for CVDLT(Computer Vision & Deep Learning Tools)?

A Python server implementing the Model Context Protocol (MCP) for image object detection, segmentation, and pose estimation using Ultralytics models (YOLOv10, YOLOv8, SAM). It is built on the MCP Python SDK and designed for integration with MCP-compatible clients.

How to use MCP Server for CVDLT(Computer Vision & Deep Learning Tools)?

Install dependencies with uv sync and download required model weights (yolov10b.pt, yolov8n-seg.pt, yolov8n-pose.pt, sam_b.pt) into the ./checkpoints directory. Start the server in stdio mode with python server.py or in SSE mode with python server.py sse [port]. For Claude Desktop, add an SSE entry in claude_desktop_config.json.

Key features of MCP Server for CVDLT(Computer Vision & Deep Learning Tools)

Detect objects in images using YOLOv10
Segment objects in images using YOLOv8
Segment entire images using Ultralytics SAM
Estimate human poses in images using YOLOv8
Support for local file paths and network image URLs
MCP tool integration with stdio and SSE transport protocols

Use cases of MCP Server for CVDLT(Computer Vision & Deep Learning Tools)

Detect and classify objects in images with bounding box output
Generate object-level segmentation masks for images
Perform whole-image segmentation using SAM
Estimate human pose keypoints from images
Integrate computer vision capabilities into MCP-based applications

FAQ from MCP Server for CVDLT(Computer Vision & Deep Learning Tools)

What model files are required?

The server requires yolov10b.pt, yolov8n-seg.pt, yolov8n-pose.pt, and sam_b.pt placed in the ./checkpoints directory.

What types of image inputs are supported?

Both local paths (using file:// or relative paths) and network URLs (using http:// or https://) are supported.

What transport protocols does the server support?

Stdio (default) and SSE (started with python server.py sse [port]).

Is there a reference client for testing?

Yes, an MCP Playground client is available at https://github.com/MRonaldo-gif/mcp-playground-local.

Where can I download the required model weights?

Download links are provided in the Ultralytics documentation for YOLOv10, YOLOv8, and SAM‑2 models.

Comments

More Productivity MCP servers

ATLAS: Task Management System

cyanheads

A Model Context Protocol (MCP) server for ATLAS, a Neo4j-powered task management system for LLM Agents - implementing a three-tier architecture (Projects, Tasks, Knowledge) to manage complex workflows. Now with Deep Research.

475