digiKam 7 Released, implemented deep learning for faces detection

July 19, the development team behind the open-source photo editing and management application “digiKam” announced the latest major release “digiKam 7.0.0”.

digiKam is an open-source photo management application. It runs on Linux, macOS, and Windows, and it offers tools to import, manage, edit, and share photos and RAW images.

digiKam 7.0 is the latest major release after version 6 series, released in February 2019.

Most notable among all the enhancements included in this release is the faces management feature. Introduced in v2.0, the faces management feature used Cascade Classifier instead of deep learning. It wasn’t powerful enough to facilitate the faces management workflow automatically. This release introduces new code, based on Deep Neural Network from the OpenCV library. As is uses neuronal networks with pre-learned data models dedicated to faces management, face detection and recognition can be used right away. Success rate is 97%, and it can also detect non-human faces such as dogs. Run-time speed is also improved.

Regarding support for RAW images, this version is based on libraw 0.20 to support over 40 new RAW formats including Canon CR3 and Sony A7R4. Support for HEIF format, which was introduced in version 6.4, is also improved.

It now supports Qt 5.15, and all the binary bundles have been switched to Qt Qt 5.14.2 LTS. It also provides official support for FlatPak.

It also introduces ImageMosaicWall as a third party plugin. The SlideShow tool has been ported as a plugin. It also includes a new set of settings that plays images in shuffle mode. As to metadata management, it adds a new option to write geolocation information into the file’s metadata.

digiKam
https://www.digikam.org