README.md

metadata

title: CV MCP Server
emoji: 💻
colorFrom: yellow
colorTo: blue
sdk: gradio
sdk_version: 6.0.1
app_file: app.py
pinned: false
license: mit
short_description: Real-time CV VLM MCP Server for MCP 1st Birthday Hackathon
tags:
  - building-mcp-track-creative

🎥 CV MCP Server

A Model Context Protocol (MCP) server that provides real-time scene analysis for webcam images.
This Space allows users to stream live video feeds and get detailed insights about the environment, objects, humans, and more.

Check out the Hackathon details here.
The social media post of this MCP Hackthon project on Discord Instagram Thread.

🎬 Watch a demo of the CV MCP Server analyzing a robot’s camera feed:
Demo Video

GitHub repo of the cv robot python script: https://github.com/OppaAI/CV_Robot_MCP

---

🌟 Key Features

• Real-time scene description
• Human detection
• Animal and object detection
• Environment classification (indoor/outdoor)
• Lighting condition analysis
• Hazards identification
• Optimized for context window efficiency

---

🔧 How It Works

This Space leverages the MCP server to analyze images captured from your webcam:

1. Stream an image from your webcam.
2. The image is sent to the MCP server.
3. The server processes it using the Model Context Protocol.
4. Outputs are returned and displayed in the UI, including:
   • General scene description
   • Detected humans
   • Detected animals and objects
   • Environment type (indoor/outdoor)
   • Lighting condition
   • Hazards

---

⚡ Demo

Compatible with PC, mobile, and robots with cameras.

Stream images via your webcam or phone camera to receive real-time scene analysis.

Watch a demo video of the CV MCP Server analyzing the video feed from my robot:
Demo Video

---

🔑 Requirements

• A valid Hugging Face API Token is required.
• If running locally, set your token as an environment variable:

export HF_TOKEN=your_huggingface_token