Digikam 8.0 Introduces Advanced AI-Based Image Classification and Support for New Formats

Digikam 8.0: New image formats and classification by AI

After two years of work and 450 fixed bugs, the free photo management software Digikam has released its main version 8.0. The software is taking its first steps towards the GUI toolkit Qt 6, which the upcoming Linux desktop KDE Plasma 6 is also based on. Digikam is still compatible with the Qt 5 libraries in its current 1.5 million lines of code, while the newer toolkit matures.

Digikam uses the Libraw library to support manufacturer-specific image formats in raw format. The software now has compatibility with cameras such as Canon EOS R3, R7, R10, Leica M11, Sony A7-IV, the Olympus OM-1 and the drone DJI Mavic 3. Digikam can thus import raw data images directly without having to use another tool. In total, the software supports around 1,220 cameras, smartphones and drones.

Digikam now offers new file types among the target formats, including JPEG-XL, WEBP, HEIF, and AVIF as output formats with or optionally without lossy compression. The software also supports database-supported data such as descriptions, annotations, copyright notices, and changed tags, and directly supports changes in the metadata of image files.

In version 8.0, Digikam can use the Exiftool as an option, which replaces the previously used Exiv2 library. Developers cite better performance and write access to meta tags in DNG and RAW files as advantages. Digikam has been experimenting with face recognition in images with neural networks since version 2.0, which recently received significant improvements with version 7.0.

Digikam 8.0 offers an optional tunable deep learning model. This model can classify and sort imported photos according to pre-trained aesthetic criteria. With an automatic evaluation, the criteria used include recording quality, content, composition, colors, and image composition with the placement of objects. This AI method allows failed snapshots to be sorted out quickly using batch processing.

Digikam 8.0 also includes an interface to the external text recognition engine of Tesseract OCR 4.1. This feature searches images or scans for writing in different languages ​​and saves the text in separate text files or in the metadata of the images themselves.

Digikam is also available for download as a port on Windows and Mac OS X. For Linux systems, Digikam 8.0 is available as a universal app image that delivers all the required Qt libraries with a static link and thus runs under desktop environments other than KDE.

The team behind Digikam is satisfied with the revised online documentation for the software, which is now made with the Sphinx tool instead of Docbook and is available as an EPUB file. The last major milestone for Digikam was version 7.0.0, which came out in mid-2020.

Leave a Reply