Restaurant Bill Reader

Table of Content:

1. Problem Statement

2. Modules Used

3. Solution Approach

4. Installation

   • Checking Wand
   • Checking Pytesseract
   • Checking OpenCV

5. Execution Instructions

6. License

7. Version

8. Author

---

Problem Statement

In this problem we are given a bunch of resturant bills in pdf format. We have to extract text from the images of bills given in ".pdf" files.

---

Modules Used

The major python modules used for solving the above mentioned problem are as follows:

• Wand
• OpenCV
• PyTesseract OCR

---

Solution Approach

1. Use Wand to convert pdf to image of any resolution (here we have used 700 x 700) and save it in images folder.

2. Read the generated image using OpenCV

3. Use PyTesseract to read text from the images and save the data obtained in a json file (here json/img-to-text.json ).

---

Installation

In this project Python version 3.7.7 is used.

First create a new anaconda environment and then activate the environment:

# Create environmemt.
conda create -n bill-reader python=3.7
# Activate environment.
conda activate bill-reader

Then install the following python packages using pip:

$ pip install wand

$ pip install pytesseract

$ pip install opencv-python

---

Checking Wand

STEP-1

Open Python terminal by typing the following command in anaconda command prompt: $ python

This will open a python terminal.

STEP-2

from wand.image import Image as wi

If you get error any error proceed to Step-3:

Visit the following link and follow the instructions given for your respective OS.

For Wndows.

Checkboxes that must be ticked while installing are as follows:

And then check again repeat Steps 1 and 2. Hopefully it will solve the import error with wand module.

STEP-3

If there is no error, then wand module is working fine. And we will exit the terminal.

quit()

STEP-4 Now open 01-pdf-to-image.ipynb file and run the cells in your jupyter-notebook.

If you get DelegateError, do the follows:

• INSTALL GHOSTSCRIPT

---

Checking pytesseract

STEP-1

Open Python terminal by typing the following command in anaconda command prompt: $ python

This will open a python terminal.

Step-2:

Visit this link and download the write installer according to your python architecture (32 or 64). Then install it and make a note of the installation location.

Then open '02-image-to-text.ipynb' file and in cell 1 update the path mensioned to your installing location.

---

Checking OpenCV

STEP-1

Open Python terminal by typing the following command in anaconda command prompt: $ python

This will open a python terminal.

STEP-2

import cv2

If you get error any error proceed to Step-3:

Refer this answer -- https://stackoverflow.com/questions/19876079/cannot-find-module-cv2-when-using-opencv

else refer this: opencv_installation_instructions

It worked for me.

---

Execution Instructions

1. First run 01-pdf-to-image.ipynb.

   It will take some time to execute completely depending upon your computer hardware.

2. Now run 02-image-to-text.ipynb.

   It will also take some time to execute.

---

License

MIT © 2020 Deepankar

---

Version

1.0.0

---

Author

The author of this project is Deepankar.