ddddocr CAPTCHA Recognition MCP Server

# ddddocr CAPTCHA Recognition MCP Server A powerful MCP server for CAPTCHA recognition using ddddocr library, providing advanced text OCR, object detection, and slider matching capabilities. ## Features - 🔤 **Text OCR Recognition** - Recognize text content from CAPTCHA images - 🎯 **Object Detection** - Detect target objects in CAPTCHA images - 🔄 **Slider Matching** - Match slider CAPTCHA positions with high accuracy - ⚡ **High Performance** - Built on ONNX runtime for fast processing - 🔌 **MCP Compatible** - Fully compatible with Model Context Protocol ## Installation & Usage ### From Smithery (Recommended) 1. Visit [Smithery.ai](https://smithery.ai) 2. Search for "ddddocr-captcha-recognition-ymeng98" 3. Install with one click to your AI toolchain ### Local Development ```bash # Clone and setup git clone <repository-url> cd ddddocr-captcha npm install # Install Python dependencies pip install -r requirements.txt # Run development server npm run dev ``` ## Tools Available ### `ocr_recognize` Recognize text content from CAPTCHA images. **Parameters:** - `image_base64` (optional): Base64 encoded image data - `image_path` (optional): Path to image file ### `detect_objects` Detect target objects in CAPTCHA images. **Parameters:** - `image_base64` (optional): Base64 encoded image data - `image_path` (optional): Path to image file ### `match_slider` Match slider CAPTCHA position. **Parameters:** - `target_base64` (optional): Target image base64 encoded - `background_base64` (optional): Background image base64 encoded - `target_path` (optional): Target image file path - `background_path` (optional): Background image file path ### `health_check` Check ddddocr service health status. ## Technical Stack - **Core Recognition**: ddddocr library - **Image Processing**: OpenCV, Pillow - **Protocol**: Model Context Protocol (MCP) - **Runtime**: TypeScript (Node.js) + Python backend - **Models**: ONNX-based neural networks ## License MIT License ## Support For issues and questions, please visit our GitHub repository or contact the maintainer.