***Above, an example of Google OCR API, hopefully, I will be able to do the same with Tesseract, one day

Installation

The installation of this library took me longer than usual.

!pip install pytesseract

While you should start by installing pytesseract using pip, if you try to run the library, it will run an error.

TesseractNotFoundError: /usr/bin/tesseract is not installed or it's not in your PATH

The installation is a little bit hectic. In fact, you will first need to install another package called tesseract-ocr and make a direct cmd connection to the .exe file (all written in the instructions and available on my repo, do not despair).

!sudo apt install tesseract-ocr

Make sure you are installing both libraries together.

Import Libraries

try:
 from PIL import Image
except ImportError:
 import Image
import cv2
import pytesseract

Checking installation folder

Before proceeding, you will need to find out where do you have to find the tesseract execution file.

!which tesseract
/usr/bin/tesseract

You can now copy the output to specify the location of the .exe file. Unfortunately, it appears this is the only workaround to make Tesseract work on Google Colab. So far, this appears to be the only working tutorial between the many I searched.

pytesseract.pytesseract.tesseract_cmd = (
    r'/usr/bin/tesseract'
)

The library should have been imported correctly.

Importing Image

I will be using the cv2 library to import and edit images. I will have to make sure that in my notebook storage I have uploaded the image I want, and that I can access its path correctly. In this case, the image is called image.png.

#deep-learning #ocr #artificial-intelligence #image-to-text #tesseract #deep learning