Brocr

A browser based OCR engine working with a python wrapper of Tesseract in the backend. The front end is rendered using Flask. The latest version of Tesseract is at play here and it gives surprisingly good outputs with a few preprocessing steps. It’s still a work in progress – I haven’t yet figured out why my trained files (.traineddata) get automatically deleted from the server save the default one (view the detailed question in stackoverflow.)

Once I figure that out, I’ll add more language supports and choices of preprocessing.
You can view the current status of the work on heroku.

Warning: This will break if you select the Bengali option from the dropdown.