Nextcloud Recognize

Nextcloud Recognize is an AI media tagging app for people organizing photo, video and music collections on a self-hosted Nextcloud server. It processes images on that server and doesn't send sensitive data to cloud recognition services.

• Face grouping. Group photos by the people who appear in them and browse those groups in the Memories app.
• Photo categories. Automatically tag animals, food, vehicles, buildings, landscapes, landmarks and monuments.
• Music and video classification. Identify music genres and human actions in videos, with tagged music available through Audioplayer and tagged photos and videos through Memories.
• Library tagging. Store recognition results in Nextcloud's Collaborative Tags and exclude selected folders from image or music recognition.

Recognize uses pretrained models including EfficientNet v2 for objects, Musicnn for music genres and MoViNet for video classification. It needs roughly 4GB of free RAM. Native processing requires an x86-64 processor with AVX and a system using glibc. A slower WebAssembly mode supports ARM processors and Alpine Linux, including Nextcloud AIO.

The app is open source under AGPL-3.0. Maintenance focuses on compatibility and critical fixes, with community contributions accepted. It can't process end-to-end encrypted files because the server can't read their contents.