13.5 OCR Systems

Optical Character Recognition (OCR) converts pictures of text, such as scanned documents or photos of signs, into machine-readable characters you can search and edit. A typical pipeline detects where the text is, then recognises the characters, and often adds a language model to fix likely mistakes from context.

A government office digitising old Amharic land records could use OCR to turn stacks of paper into searchable files. Because Amharic uses the Fidel script with hundreds of characters, the model must be trained on Amharic specifically, not just Latin letters.