PhotoPrism

PhotoPrism is a self-hosted photo management application that uses TensorFlow to automatically classify and tag photos by content, location, and subject. Import your library and it generates searchable subject labels — sunsets, dogs, architecture, food — without you manually tagging anything. EXIF data from camera GPS feeds an interactive world map that shows where photos were taken. Combined with timeline browsing and automatic album suggestions, it provides a self-hosted experience that covers most of what Google Photos does.

RAW format handling is a practical differentiator for photographers. PhotoPrism supports RAW files from Canon, Nikon, Sony, Fujifilm, and most major camera manufacturers, converting and indexing them alongside JPEGs. The originals are preserved untouched; PhotoPrism works from sidecar files and converted versions.

The pricing structure is worth understanding before you set it up. The core application is open source under AGPL-3.0, but some features — including multiple user accounts, unlimited RAW conversion, and priority support — require a paid membership called Essentials or Plus. For personal single-user use the free tier covers a lot. For household sharing or team use you will likely want to check the current membership tiers.

Resource requirements are higher than lightweight alternatives like Navidrome. The initial indexing pass on a large library is CPU-intensive, and face recognition adds to that. On a capable home server or NUC this is manageable, but on a low-power Raspberry Pi or entry-level NAS the indexing process can run for days on a large library.

Immich is worth comparing if your use case is mobile photo backup with a Google Photos-style interface. PhotoPrism is stronger on browsing and classification for an existing large archive.

PhotoPrism: Pros & Cons