FOSSPicks

Photo management

digiKam 7

One of the biggest compromises you make when you forgo the privacy infringement of online photo storage is convenience. Not only is online storage usually limitless and cheap, it's typically combined with the same clever AI-generated search metadata that drives whatever online platform you choose. Google, Apple and Amazon, for example, all silently trawl through your uploaded photos to recognize common items, colors, locations, situations, and of course, faces. This can really help you find a single image in a collection that might have started over a decade ago, and it's a process that can't be easily replicated on a local machine with local storage and no access to the cloud. Until now.

DigiKam 7 is the latest release of KDE Plasma's flagship photo manager, a photo manager that also happens to work brilliantly on other desktops and (keep it quiet!) Windows and macOS. Its best new feature is one that finally delivers on a function the app has long been experimenting with: deep-learning-powered face recognition. This is thanks to contributions from Google Summer of Code developers, who first demonstrated that neural networks could improve recognition, before integrating the new deep neural network features now found in the wonderful OpenCV library. It uses a previously trained dataset to discover faces in your photos from just a single example, without any online communication. Faces can then be tagged with whatever name you assign to them. Just like any cloud service, the tags will even show the face as a thumbnail, making it trivially easy to find photos with the same person.

If you've not installed digiKam for a while, a first-run wizard steps you through the database back end (SQLite) before asking where your photos are located and whether you want metadata in the database or photos themselves. Although digiKam has always supported a huge number of image formats, this release adds support for Canon's new CR3 RAW files. This is an important addition when you consider how prevalent Canon cameras are in the open source community, thanks to the fabulous Magic Lantern firmware. One of the best things about digiKam is that despite the huge size of many RAW files the import, navigation, and loading of photos is typically very fast. Facial recognition can slow this down, but fortunately you can turn on an option to scan new images automatically.

[...]