We process a lot of supplier invoices and shipping docs that come in Spanish, Portuguese, and sometimes Mandarin. The OCR seems to handle Spanish fine but I’m curious about other languages. Anyone tried multilingual setups?
2 Likes
We get some documents in Hindi and Gujarati. The OCR accuracy drops a bit compared to English but it’s still usable for structured documents like lab reports. For handwritten stuff it struggles though.
2 Likes
In my testing, CJK documents work reasonably well if they’re typed/printed. The key thing is making sure your template field mappings account for different text directions and character sets. I had to adjust some regex patterns in my post-processing pipeline.
2 Likes