immich-machine-learning-cpu ghcr.io/immich-app/immich-machine-learning:release https://github.com/immich-app/immich/pkgs/container/immich-machine-learning immich_internal bash false https://github.com/immich-app/immich/issues https://immich.app Immich Machine Learning service. Handles face recognition, CLIP-based image search, and OCR. This service runs ML models locally. Models are downloaded on first use and cached. The default cache can grow to several GB depending on enabled features. For GPU acceleration, change the image tag to: release-cuda, release-rocm, release-openvino, release-armnn, or release-rknn. Must be on the same Docker network as the Immich server. MediaApp:Photos AI: Status:Stable https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/immich-machine-learning.xml https://raw.githubusercontent.com/rorar/unraid-templates/main/templates/docker_icons/immich-ml-logo.png This template uses your CPU for machine learning. For hardware accelerated ML, use one of the template that fits your (i)GPU. Must be on the same Docker network as the immich-server container. The server connects to this service on port 3003 by container name. Full setup guide: https://rorar.github.io/immich-unraid-manual/ /mnt/user/appdata/immich/model-cache/ Europe/Berlin 300 1 120 log