OCR service

The OCR service consists of a backend API, which is a third-party written Rust api, which enables sending images up to it, and choosing a Tesseract model to read the text out of the image.

We run Nasjonalbiblioteket’s models, as well as our own. The ones from NB generally produces more accurate readings.

Accompanying the backend API, there is also a simple frontend application, written in SvelteKit.