PhotoPrism AI Photo Library
PhotoPrism is an AI photo library written in Go. Its pitch is simple: point it at a messy folder of photos and let it recognise people, places and content for you. It does not try to own phone backup, but it is excellent at organising, searching, mapping and RAW handling — ideal for photographers and families with years of unsorted archives.
What it does well
- Automatic classification: TensorFlow-based labels, scenes and faces; browse by person, place, colour or year.
- Powerful search: keywords, filters, similar images and geo-fencing, fast even with tens of thousands of photos.
- Wide format support: JPEG, PNG, HEIC, RAW (CR2/NEF/ARW and more) and video, with originals left untouched.
- Single container: SQLite by default, so a small 2 vCPU / 4 GB box handles tens of thousands of photos.
Images and ports
Official image photoprism/photoprism:latest on Docker Hub, web port 2342. Switch to MariaDB only when the library or concurrency grows.
Deployment cost
2 vCPU / 2 GB is the floor, 4 GB is comfortable. First indexing is CPU-heavy, so run it overnight. Mount three directories: originals, import and storage — and do not forget storage when backing up.
Full walkthrough: PhotoPrism Docker deployment tutorial.
Having trouble setting up this project?
We have run “PhotoPrism AI Photo Library” locally and verified that it installs and the admin panel works. We can deploy it to your server.
- Starting at
- ¥99+
- Estimated time
- 48 hours
- Deployment notes
- 目录规划、Compose 部署、首次索引、HTTPS 反代 / Volumes, Compose, first index, HTTPS proxy
WeChat: ITcloub · Bring your own server and domain · Full refund if deployment fails