when uploading photos and videos in google pictures, it will automatically categorize the picture in places and things (like Skylines, Temples, Mountains, Lakes). I really like this feature. I know google is a big player when it comes to AI things like that, but is there any way to get something like this with nextcloud?
You would have to put your own trained AI there. Using external services could be done but normally you host data yourself just to not do it.
Have a look at
As you can see, there was some activity but nothing is installable in a simple way right now.
Matias and me are also working on self-hosted AI-powered face recognition (and face clustering) app. It is mostly ready, it is just not in app store yet. Take a look, this might be interesting to you. Here is me talking about it here - Search images for content of image, and here is GitHub - https://github.com/matiasdelellis/facerecognition.
Now, for your needs, there are two possible ways for you to continue (or anyone reading this thread who wants to get his/her feet wet:grinning:)
- Since our app is supposed to be used as a core app that delivers value (about faces) to other apps (have API to other apps), you can use it in another/your app, to auto-tag images that contain/do not contain faces (or even specific persons! or persons from Contacts, or from CarDav, go wild!).
- Since it is using powerful and mature dlib library underneath, we spent a awful lot of time lowering friction to users installing this - preparing dlib for us, integrating dlib with php, creating packages for distributions, making sure it works with or without GPU, work in background process with a queue…you can take our application as a starting template, create your own, similar in architecture. You will need to find/create some model for DLib with lakes, mountains…
Ping us on github for more information!
Hope it helps!