5

I have a Python app which words great via Localhost on my machine.

I am trying to deploy it to Heroku. However it does not seem possible to accomplish this (I have spent approx 30 hours trying now).

The problem is Tesseract OCR. I am using the pytesseract wrapper, and my code utilises this. However, no matter what I try, it does not seem to be possible to use pytesseract when it is uploaded to Heroku.

Could anyone either suggest how to go about deploying a Hello World Tesseract OCR Python app via pytesseract to Heroku, or if Heroku is not capable of this, suggest an alternative to Heroku?

user3795126
  • 109
  • 2
  • 5

2 Answers2

11

For anyone else looking to deploy pytesseract on heroku, here are the steps:

  • Add apt buildpack to heroku 

    heroku buildpacks:add --index 1 https://github.com/heroku/heroku-buildpack-apt

  • Create a file named Aptfile in the same directory as your app and these lines to it (for english)

    tesseract-ocr
tesseract-ocr-eng

  • Set heroku config variable TESSDATA_PREFIX to the path of tessdata folder (it was /app/.apt/usr/share/tesseract-ocr/4.00/tessdata for me) which can be found out by running heroku shell using heroku run bash and run this command in the shell

    find -iname tessdata

  • Set the config variable using

    heroku config:set TESSDATA_PREFIX=/app/.apt/usr/share/tesseract-ocr/4.00/tessdata

    replace the path with the path you got from the previous command

Tesseract should be installed in the heroku app when you build it. Make sure you have pytesseract in your requirements.txt file. Now you should be able import and use pytesseract on heroku

Samyak S Sarnayak
  • 113
  • 3
  • 6
  • For me, on Mac, the path was empty. I tried with your path only, and it worked for me. ;-) – Kishor Pawar Feb 03 '20 at 13:01
  • @KishorPawar you should run the find command in `heroku run bash` and not on your local machine. Also make sure that you push the changes to Heroku before running that command. – Samyak S Sarnayak Feb 04 '20 at 15:14
  • 1
    Yes, I ran the command on `heroku bash` only. But anyway, your path worked for me. ;-) – Kishor Pawar Feb 04 '20 at 15:17
  • Hi, do we have heroku 18 compatible buildback for tesseract?..however this buildpack is not working for me..says its not compatible. – Jayahe Sep 22 '20 at 04:22
  • @SamyakSSarnayak Hello. Do you know how to install specific version of Tesseract-ocr (on the Aptfile), instead of the 4.0.0? – Rohalt Feb 15 '21 at 03:46
0

The Python buildpack likely doesn't have the tesseract binaries installed by default. Here’s a third-party buildpack that creates the appropriate environment. Follow the instructions there to make the binary available to your application.

RangerRanger
  • 2,455
  • 2
  • 16
  • 36
  • Thanks for the suggestion! In the end, In the meantime I decided to try the Google Cloud service instead of Heroku, and with some assistance managed to get it working there instead. :) – user3795126 Dec 08 '18 at 12:32