digiKam 8.3 can now automatically tag photos using a deep-learning engine

digiKam, the professional photo management software, has announced the release of its version 8.3. The highlight of this update is automatic tagging based on contents, courtesy of a deep-learning engine with a pre-trained model. This tool can detect various forms, objects, places, animals, plants, monuments, scenes, and more, and generates a series of keywords parented with the Auto branch in the database. User review is then required to validate the detected contents.

The latest version also introduces features to support audio tracks, including encoding time-lapse video with frame luminosity equalization and On Screen Display support for metadata such as date, comments, aperture, lens properties, etc.

The Video-SlideShow plugin has seen improvements as well. The encoder, previously reliant on QtAV for encoding and rendering video frames from images, has been ported to the FFMpeg CLI tool and QtAVPlayer for rendering the result.

digiKam 8.3 brings an enhanced Windows version, now built on Qt and KDE frameworks version 6. The macOS and Linux bundles continue to run on Qt version 5, but a migration to version 6 is slated for later this year.

This version updates the internal RAW decoder Libraw to the rolling-release snapshot 2024-02-02. It also marks the completion of the transition to the QtMultimedia with Qt6 framework, while the Qt5 version now uses a new framework named QtAVPlayer.

digiKam and Showfoto now support 61 different languages for the graphical interface, indicating an update in application internationalization. With digiKam 8.3, longstanding bugs from previous versions have been fixed, alongside other improvements and add-ons.