Testing Layout Parser on medieval Canon Law texts

The Data Science Center gets many requests for OCR and usually reaches for ABBYY FineReader or Tesseract. DataSquad members tested Layout Parser, a newer collection of OCR packages that is good at separating a page into boxed sections, the way a menu separates mains from drinks.

They tried it on Canon Law, a set of medieval texts in UCLA’s digital collections. Layout Parser separated the sections well, but it was hard to set up. The team got it running inside a container configured with the exact system requirements the software needs.

Their conclusion was that the installation difficulty and the lack of support for the software are major disadvantages. That kind of recorded negative result is useful: the next person asked about this tool can read the answer instead of repeating the afternoon.