EasyOCR

You can run a self-contained MCP server that provides OCR capabilities powered by EasyOCR. This server exposes simple tools to process images from different sources and return either plain text or detailed results with coordinates and confidence, making it practical for automating document analysis, form processing, and multilingual text extraction.

How to use

To use the EasyOCR MCP Server from your MCP client, connect to the local or remote MCP endpoint and call the available OCR tools. You can process images from base64 strings, files on disk, or image URLs. Choose the output detail level to get either just the extracted text or detailed results that include bounding box coordinates and confidence scores.

Key tools you’ll interact with are ocr_image_base64, ocr_image_file, and ocr_image_url. Each tool returns results in one of two detail modes. In detail level 0 you receive a list of text entries. In detail level 1 you receive a list of text segments with their bounding boxes and confidence scores.

How to install

Prerequisites you need before running the server: Python environment and a runtime capable of executing MCP servers. You also need the dependency stack for EasyOCR, which can use GPU acceleration if you installed PyTorch with CUDA support.

Install PyTorch with GPU support if you plan to use GPU acceleration. If you will run on CPU only, you can skip this step.

Install all dependencies and set up the environment, then run tests to verify the implementation.

uv pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

uv sync

uv run test.py
uv run test-gpu.py

Configuration and running

The server runs as an MCP process that you start from your MCP client setup. You can start it using the provided runtime command as part of your MCP configuration.

You can configure the MCP server through a parent MCP configuration file. The following shows how to define the EasyOCR MCP server using an HTTP-based launcher and, alternatively, a local stdio launcher.

// Windows example using the standard MCP launcher
{
  "mcpServers": {
    "easyocr-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "X:\\path\\to\\your\\project\\easyocr-mcp",
        "run",
        "easyocr-mcp.py"
      ],
      "env": {
        "EASYOCR_LANGUAGES": "en,ch_tra,ja"
      }
    }
  }
}

// Linux/macOS example using the standard MCP launcher
{
  "mcpServers": {
    "easyocr-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/your/project/easyocr-mcp",
        "run",
        "easyocr-mcp.py"
      ],
      "env": {
        "EASYOCR_LANGUAGES": "en,ch_tra,ja"
      }
    }
  }
}

Available tools

ocr_image_base64

Process a base64-encoded image and return OCR results. You can choose detail level to get text only or full coordinates with confidence.

ocr_image_file

Process an image file from disk and return OCR results. You can configure detail level and merging thresholds.

ocr_image_url

Process an image from a URL and return OCR results. You can configure detail level and paragraph detection.