Florence-2

An MCP server for processing images using Florence-2

python

6

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "jkawamoto-mcp-florence2": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/jkawamoto/mcp-florence2",
        "mcp-florence2"
      ]
    }
  }
}

Florence-2 MCP Server processes images or PDFs to extract text via OCR or to generate descriptive captions that summarize image content. It is designed to run as a local MCP server that you connect to with your MCP client, enabling you to offload image understanding tasks to Florence-2 from your data sources.

How to use

You connect to the Florence-2 MCP Server from an MCP client to send images or PDFs for OCR or captioning. Choose OCR when you need extracted text from a source image or document page, or choose captioning to obtain natural language descriptions of visual content. You can operate with locally stored files or fetch files from web locations, then receive the processed text or captions as the response.

How to install

Prerequisites you need before starting: a system with a supported operating system and a runtime package that can run the MCP server commands described below.

Step 1. Prepare the MCP entry for Florence-2 in your client’s configuration. Use the provided stdio configuration that runs the MCP server locally.

Step 2. Start the MCP server using the command shown in the configuration example.

Additional configuration and usage notes

The Florence-2 MCP Server can be integrated with multiple MCP clients. For Claude Desktop, you configure a local MCP server entry that points to the Florence-2 project and uses the uvx runner to launch it. Goose and LM Studio offer similar pathways to connect and configure the server.

Tools

ocr: Process an image file or URL using OCR to extract text.
caption: Process an image file and generate captions describing the image.

Configuration and startup example

{
  "mcpServers": {
    "florence2": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/jkawamoto/mcp-florence2",
        "mcp-florence2"
      ]
    }
  }
}