This server wraps automatic license plate recognition into both REST and MCP interfaces, letting you detect and read plates from images without embedding Python ML dependencies into your application. It exposes tools like detect_and_recognize_plate_from_path that take image URLs or local paths, plus variants that work with base64 encoded data or file uploads. The recognition pipeline handles both detection and OCR, with support for CPU (ONNX), Intel OpenVINO, and CUDA acceleration. You'd reach for this when building applications that need ALPR capabilities but want to keep the ML stack isolated in a separate service, or when connecting LLMs to computer vision tasks through the MCP protocol.
claude mcp add --transport stdio habedi-omni-lpr uvx omni-lpr