Ask HN: What software would you like to see open source/free versions of

Re:Ask HN: What software would you like to see open source/f... by staticautomatic on Monday October 23, 2017 @06:56PM (#15537611)

Zonal OCR software that can do data extraction as well as the stupid expensive offerings from ABBYY and Nuance.

[ Parent | on HN ]

Re:Ask HN: What software would you like to see open source/f... by ocrcustomserver on Monday October 23, 2017 @09:04PM (#15538348)

Can you elaborate? What is the use case?

[ Parent | on HN ]

Re:Ask HN: What software would you like to see open source/f... by staticautomatic on Tuesday October 24, 2017 @02:04PM (#15543726)

Extraction from semi-structured documents. Reqs:

- No absolute element positions

- Able to handle a reasonable amount of skew

- Able to handle photographs of documents (e.g. mobile)

- Excellent support for repeating groups/elements

- Around 95% accuracy without human verification

[ Parent | on HN ]

Re:Ask HN: What software would you like to see open source/f... by loxias on Saturday October 28, 2017 @04:19AM (#15574425)

Similar to the other comment you received, I've also been working on some of the math necessary to do such a a thing! I have no idea how it would be productized there's so many things out there that do "good enough" jobs... One of my ideas was to extract structure from motion and use that to deskew and align..

(Can I also have a datadump of candidate files for testing?

[ Parent | on HN ]

Re:Ask HN: What software would you like to see open source/f... by ocrcustomserver on Tuesday October 24, 2017 @06:42PM (#15545838)

Interesting, I have implemented some version of the above in the past (albeit not in an opensource/free solution).

If you have a specific document/data extraction need in mind, you can email me a sample and I will let you know if I can help.

[ Parent | on HN ]