Immich-machine-learning

Machine-learning service for Immich (face recognition, smart search, etc).

  • AI
  • Media Applications
  • Other
  • Tools & Utilities
Machine-learning service for Immich (face recognition, smart search, etc). This is the "immich-machine-learning" service from the official Immich docker-compose.yml. It must be on the same custom Docker network ("immich") as the Immich server, which reaches it at http://immich-machine-learning:3003 via the IMMICH_MACHINE_LEARNING_URL setting. It exposes port 3003 only on the internal network (no host port published). For hardware-accelerated inference, switch the repository tag to one of: release-armnn, release-cuda, release-rocm, release-openvino, release-rknn, and add the relevant device(s) - see https://docs.immich.app/features/ml-hardware-acceleration.

Platform availability

Available on 1 of 10 platforms

  • Unraid
  • TrueNAS (not listed)
  • Umbrel (not listed)
  • ZimaOS (not listed)
  • Proxmox (not listed)
  • Helm (not listed)
  • Mac (not listed)
  • Windows (not listed)
  • Linux (not listed)
  • Any server (not listed)

Health score?A 0-100 score computed nightly from four factors: maintained (40%: recent commits and steady activity), popular (25%: GitHub stars, log scale), easy to install (20%: platforms, install notes, setup guides, Docker image) and light to run (15%: minimum RAM, ARM support). When the source code has been checked, a fifth factor, security check (20%: automated scan for committed secrets, unsafe desktop-app settings, known-vulnerable dependencies and project hygiene; any serious finding caps it at 20), is added. Factors without data are left out and the rest are rescaled.

74/100 Good

Maintained
100last commit 17 days ago
Popular
100118,238 GitHub stars
Easy to install
351 install platform, install notes, Docker image
Light to run
106 GB minimum RAM

Availability: platform listings confirmed today, links checked 21 days ago

GitHub stars
118,238
Open issues
711
Last commit
2026-11-04
Latest release
v3.3.6
License
AGPL-3.0
Activity
steady

Checked 17 days ago - source: GitHub (immich-app/immich)

Resources & Compatibility

Minimum requirements: 6 GB RAM, 2 CPU core(s).

Checked 54 days ago - source: Docker Hub image tags and published docs

First-install notes

  • Put the config where Docker expects it Download the example config.yaml into your ./config folder before running the container. The kiosk reads all of your settings from that file, so without it you won't have a working setup. source
  • Fill in Immich URL and API key Edit config.yaml and set immich_url and immich_api_key to point at your Immich server. If these are blank or wrong, the kiosk can't log in or show your photos. source
  • Launch the slideshow in a kiosk app On the TV or frame, install and start either ImmichFrame or Fully Kiosk Browser, then open your kiosk URL. This keeps the photo display fullscreen and running like a frame instead of a normal web page. source
  • Hardware acceleration For GPU/CPU acceleration, change the image tag to release-cuda, release-rocm, release-openvino, release-armnn, or release-rknn and pass the relevant device(s) into the container. source

Fetched 47 days ago - each note links to its source

Alternatives (12)

  • ChronoFrame - ChronoFrame is a self-hosted photo gallery for homelab users who want a private, low-maintenance place to upload, organize, and share perso…
  • Ente Photos - Open-source, end-to-end encrypted backup and sharing for photos and videos, with multi-platform sync, original quality, privacy controls, a…
  • Frameleaf Manager - Frameleaf Manager is the guided installer and upgrade tool for the self-hosted Frameleaf photo and video library.
  • Go-vod - Go-vod is a minimal, self-hosted video transcoding server for the Memories Nextcloud photo app that exposes stored media as HTTP Live Strea…
  • Google Photos - Cloud storage offering 15 GB free space, automatic album and duplicate organization, AI editing, cross-platform sharing, and secure backups.
  • HomeGallery - HomeGallery is a self-hosted, open-source web gallery for browsing personal photos and videos while keeping your original file and folder s…
  • PhotoPrism - PhotoPrism is a self-hosted, AI-powered photo and video manager.
  • Piwigo - Piwigo is a self-hosted, open-source photo gallery for the web that lets you publish, organise, and share your picture collection with powe…
  • Proton Drive - Emphasizes privacy with end-to-end encryption, Proton Drive safely stores, synchronizes, and shares files across devices via web and mobile…
  • TinyCld - Open source suite for mail, calendar, drive, and contacts with self-hosting, privacy, zero cost, open standards, versioned files, and unifi…
  • foldergram - Foldergram is a self-hosted, local-first web gallery that scans folders for photos and videos and displays them in a fast, Instagram-style…
  • librephotos - LibrePhotos is a self-hosted, open-source photo management service for organizing and browsing photo and video collections.

Checked 54 days ago - source: AlternativeTo

Common questions

How do I enable GPU acceleration for Immich machine learning?

Change the machine-learning image tag to a backend-specific release tag, such as release-cuda or release-rocm, and pass the relevant device into the container. The feature is still experimental and may not work on all systems.

Why might the machine-learning container fail to start with a libGL error?

It may fail to start because the libGL.so.1 library is missing.

What should I do if machine learning uses too much CPU or memory?

You can mitigate this, host the machine-learning container on a more powerful system, or disable machine learning entirely.

What backup approach does Immich recommend for photos and videos?

Follow a 3-2-1 backup plan for your photos and videos.

Answers sourced from docs.immich.app, docs.immich.app, forum.cloudron.io, github.com

Community

Collected 48 days ago - public community threads