OpenCV-Sudoku-Solver

Project Demo:

https://youtu.be/vm5mjRRbf_k

What is it?

A Flask/React application that takes an image of an unsolved sudoku puzzle as input and outputs a valid solution.

Languages/Frameworks used:

1. Python
2. TypeScript
3. HTML/CSS
4. Flask
5. React
6. Material-UI
7. Makefile

Key features:

1. The user uploads an image of an unsolved sudoku puzzle and clicks 'solve' when uploaded.
2. The program outputs 'invalid puzzle' if the image is not a readable sudoku puzzle.
3. The program uses OpenCV and a custom OCR machine learning model to read the image.
4. Afterwards, the program solves the puzzle using 3 optimized logical deduction/backtracking algorithms.
5. The solution is then displayed on the frontend via React, also identifying puzzles with multiple solutions.
6. The user also has the option of manually filling a Sudoku Grid by clicking on 'Fill in the grid instead'.

Screenshots of Project

Frontend Environment variables

Variable	Description
NODE_ENV	development

Note: Frontend Environment variables can be supplied via a .env file in the client directory. See ./client/example.env for an example

Project Setup

To run the project, you will need to open two separate terminals.

• In the first terminal, run make backend. This will install all the Python packages in a virtual environment and run the server on port 5000.
• In the second terminal, run make frontend. This will install all the TypeScript packages and run the application on port 3000. Open http://localhost:3000 to view the result in your browser.

Other Make Commands

Command	Description
make install_frontend	Install all the frontend TypeScript packages
make start_frontend	Start the frontend without installing packages
make install_backend	Setup virtual environment and install backend Python packages
make start_backend	Start the server without installing packages
make clean_backend	Destroys the virtual environment and removes Python packages