MCP Openvision

🚀 MCP Open Source Project

MCP is an open-source project that aims to provide a powerful image analysis tool. It allows users to conduct detailed image analysis through simple API calls, supporting multiple input formats and advanced customization options.

🚀 Quick Start

MCP is an open-source initiative designed to offer a robust image analysis tool. With straightforward API calls, users can perform in-depth image analysis, benefiting from support for multiple input formats and advanced customization features.

✨ Features

• Provide a powerful image analysis tool.
• Support multiple input formats and advanced customization options.
• Allow users to perform detailed image analysis through simple API calls.

📦 Installation

Install Dependencies

pip install mcp-openvision

💻 Usage Examples

Basic Usage

# Analyze an image from a URL
result = await image_analysis(
    image="https://example.com/image.jpg",
    query="Describe this image"
)

# Analyze an image from a local file
result = await image_analysis(
    image="path/to/local/image.jpg",
    query="Identify all traffic signs in the street scene and explain their meaning to drivers"
)

Advanced Usage

Base64-Encoded Input

# Use Base64-encoded image data
result = await image_analysis(
    image="SGVsbG8gV29ybGQ=...",  # Base64 data
    query="Analyze this product packaging design, highlighting areas that can be improved for better visibility and brand recognition"
)

Custom System Prompts

# Add a custom system prompt for professional analysis
result = await image_analysis(
    image="path/to/local/image.jpg",
    query="Analyze the composition and artistic techniques of this painting, focusing on how they create an emotional impact",
    system_prompt="You are a professional art historian with in-depth knowledge of painting techniques. Focus on formal analysis, paying attention to composition, color, brushstrokes, and style elements."
)

Input Types

The image_analysis tool supports the following image input formats:

1. Base64-encoded strings
2. URL links - Must start with http:// or https://
3. File paths:
   • Absolute paths: Full paths starting from the root directory (Unix) or drive letter (Windows).
   • Relative paths: Paths relative to the current working directory.
   • Relative paths with a project root: Use the project_root parameter to specify a base directory.

Using Relative Paths

When using relative file paths, you have two options:

1. The path must be relative to the current working directory where the server is running.
2. You can specify the project_root parameter:

# Example: Using a relative path and the project_root parameter
result = await image_analysis(
    image="examples/image.jpg",
    project_root="/path/to/your/project",
    query="What is contained in this image?"
)

This is useful for applications with unpredictable working directories or when you want to use file references relative to a specific directory.

🔧 Technical Details

Initialize the Development Environment

# Clone the repository
git clone https://github.com/modelcontextprotocol/mcp-openvision.git
cd mcp-openvision

# Install development dependencies
pip install -e ".[dev]"

Code Formatting

This project uses Black for automatic code formatting. Formatting is enforced in GitHub Actions:

• All code committed to the repository will be automatically formatted with Black.
• Pull request (PR) branches from forked repositories will have the formatted code directly committed to the PR branch.
• PRs from external repositories will retain their original format.

Testing

# Run the test suite
python -m pytest tests/

🚀 Release Process

Package the Project

# Create a source distribution
python setup.py sdist bdist_wheel

# Install and test the publishing tool
pip install twine
twine check dist/*

# Publish to PyPI
twine upload --repository-url https://upload.pypi.org/username/ dist/*

📄 License

Sponsorship

The project accepts sponsorship from individuals and enterprises to support continuous development and maintenance.

License

The MCP open-source project is licensed under the MIT License. Click here to view the full license text.

A Word to Developers

If you encounter any issues or have suggestions for improvement while using MCP, please feel free to submit issues or provide feedback via GitHub Issues!