Immich
Self-hosted Google Photos replacement with native iOS and Android backup apps, CLIP semantic search, facial recognition, and a timeline layout that closely mirrors the cloud experience
Quick Start
wget -O docker-compose.yml https://github.com/immich-app/immich/releases/latest/download/docker-compose.yml && wget -O .env https://github.com/immich-app/immich/releases/latest/download/example.env && docker compose up -d Overview
Immich is a self-hosted photo and video library that replaces Google Photos or iCloud on your own hardware. The iOS and Android apps handle automatic background backup, so once it is running, photos and videos from your phone arrive on your server without any manual step and without leaving your network.
The part that separates Immich from simpler photo backup tools is search. It uses CLIP-based semantic search, which means you can describe a photo in plain language and it finds matches without tags or manual organisation. Type “red coat in the snow” and it surfaces the right photos. Facial recognition groups shots by person automatically. OCR reads text from screenshots and documents, making them searchable too. Beyond search: a map view for geotagged photos, shared albums, partner sharing, and a timeline layout that closely mirrors Google Photos.
Funding and longevity are less of a concern here than with most self-hosted projects. Immich is backed by FUTO, an organisation focused on sustainable free software development. The project is free to use, with an optional product key purchase available if you want to support it financially. At 102k GitHub stars, it is one of the most widely deployed self-hosted tools in any category.
Hardware requirements are worth checking before you start. The minimum is 6GB of RAM and 2 CPU cores. Machine learning features push those requirements higher in practice. On a low-powered NAS or a small VPS, facial recognition and smart search will run slowly unless you configure them to process on a schedule rather than in real time.
Immich: Pros & Cons
| Pros (The Wins) | Cons (The Friction) |
|---|---|
| Mobile backup: iOS and Android apps handle automatic background sync. | Hardware floor: 6GB RAM minimum; ML features need meaningfully more. |
| Smart search: CLIP finds photos from plain-language descriptions; no tagging needed. | AGPL-3.0: Commercial use restricted; check terms before embedding. |
| Facial recognition: Groups photos by person automatically across the full library. | Slow ML on weak hardware: Face recognition needs a decent CPU or GPU to feel responsive. |
| 102k stars, FUTO-backed: Institutional funding; one of the most deployed self-hosted tools anywhere. | Breaking upgrades: Rapid dev pace; major versions occasionally need migration steps. |
Use Cases
Specific ways to use Immich for your workflow.
Deployment Strategy
Recommended ways to host Immich in your own environment.