Introduction

Face Recognition that implements a basic facial recognition system integrated directly into Nextcloud cloud solution. This app is beta quality released under GNU Affero GPLv3 licence. We encourage everyone to play with it, tell us what you think and to generate more ideas how it can be further leveraged and integrated into Nextcloud ecosystem!

Goal of this application is face detection and face extraction from user images, analysis of those faces and their clustering into groups of persons. On top of these information, Face Recognition application is providing additional functionalities, all with built-in privacy of Nextcloud. You can think of it as Google Photos, but for faces only (not detecting objects…) and in such way that your images never leave your Nextcloud instance:smile:

Image is worth thousand words:

Head over to requirements & limitations to see if your Nextcloud machine is powerful enough to drive underlying deep neural networks, or jump straight to installation. Take a look at usage if you want to see if Face Recognition app will fulfill your expectation before installing it.

While face recognition is not in the realm of science fiction anymore, there are still very few open-source solutions that are production ready. Face Recognition app utilizes DLib which contains state of the art deep neural networks and precomputed models, for both face detection, face analysis and clustering.

Since these tasks are CPU heavy, application listens for the creation of new image files, and queues them for later analysis. A scheduled task (or administrator on demand) take this queue, and analyze these images with one simple command. If you want to take a deep dive into gory details, we have full Functional spec for more details.