Florence-2 MCP Server

Example Use Cases

• 1

  Extract searchable text from scanned documents, images, or PDFs hosted locally or on the web.

• 2

  Generate concise descriptive captions or summaries of image content to aid downstream reasoning or to present visual context.

• 3

  A drop-in MCP server to add image understanding capabilities to Claude Desktop, Goose CLI/Desktop, or LM Studio workflows.

Description

This project provides an MCP (Model Context Protocol) server that uses Microsoft Florence-2 (large) to extract text from images/PDFs via OCR and to generate descriptive captions summarizing image content. It can process files located locally or on the web and is designed for easy integration into MCP-compatible clients like Claude Desktop, Goose (CLI/Desktop), and LM Studio. The server is open-source (MIT) and includes usage tooling for two primary operations: ocr and caption.