Machine Learning in Linux: PhotoPrism – AI-Powered Photos App

In Operation

We’re not using a personal photo collection to illustrate PhotoPrism. Instead, we’ll access a licensed collection of photos which has missing metadata and a host of other issues. But for illustrative purposes it’s fine. And it avoids revealing embarrassing holiday snaps 🙂

We chose to index our images directly which leaves all file and folder names unchanged. Alternatively, there’s an import feature which automatically removes duplicates, gives files a unique name, and sorts them by year and month.

We’re not going to cover all the features of PhotoPrism but instead focus on a few highlights.

The first thing to note is that the web interface is extremely polished and well organized. A lot of thought has gone into the design. We really didn’t need to read the user documentation in this regard. Excellent stuff!

One of the features of any self-respecting photo gallery app must possess are good search filters. PhotoPrism really excels in this areas.

Pictures are automatically classified based on their content and location, and you can search other image properties too. Combine that with automatic classification of pictures based on their content and location and you’ll be finding the pictures you want quicker than ever!

PhotoPrism supports indexing, viewing, and converting most popular image, video and RAW formats, including JPEG, PNG, GIF, BMP, HEIF, HEIC, MP4, MOV, WebP, and WebM.

Original media and sidecar files are scanned for Exif and XMP data as well as proprietary metadata, including Google Photos JSON. The combined information is normalized and merged.

Next page: Page 3 – Facial Recognition

Pages in this article:
Page 1 – Introduction / Installation
Page 2 – In Operation
Page 3 – Facial Recognition
Page 4 – Places
Page 5 – Summary