Hacker News Reader

This is great. Did you tune the recognition models at all or is it straight tesseract plus the wrapper.

The aim of the project was to bring most tesseract's API functionality to a pythonic interface without the need for any subprocess calls. No model tuning involved, just a wrapper :)

Great work, btw!

Feature request: It would be great if you could also hook-in to the model training API, if any, of Tesseract. That way, you could drive the training via python rather than using Tesseract's accompanying executables.

It's possible the required methods are already there but I'll look into it. Would be cool if you open an issue on GitHub. Thanks for the feedback :)

Reply on news.ycombinator.com