LibrePhotos/librephotos

What it solves

LibrePhotos provides a self-hosted, private alternative to cloud-based photo management services. It solves the problem of organizing large, scattered photo collections by automatically categorizing images, identifying people, and enabling natural language search without relying on third-party cloud providers.

How it works

It uses a backend built on Django that runs local machine learning pipelines to process your images. It employs InsightFace for face detection and clustering, OpenCLIP for generating image tags and semantic search capabilities, and ONNX Runtime to execute these models efficiently. The system scans your existing photo directories, extracts metadata, generates captions, and indexes everything into a PostgreSQL database, which is then served through a React-based web interface and a mobile app.

Who it’s for

This project is for privacy-conscious users who want to manage, search, and organize their personal photo and video libraries locally on their own hardware, including support for Docker, Kubernetes, and Unraid environments.

Highlights

• Automatic face recognition and clustering for easy people-tagging.
• Semantic search functionality (e.g., searching for "dog on a beach") powered by CLIP.
• Support for diverse file formats including RAW, HEIC, and JPEG XL.
• Privacy-focused architecture with no reliance on external cloud AI services.
• Cross-platform accessibility via a web interface and dedicated mobile apps for Android and iOS.