Real life Sudoku Solver

The main language used is python and the libereries used are OPEN CV and Tensorflow 2

Input clicked from TOI IMG:

• the main couse of this project is to find a to solve a sudoku with just a clear photo of sudoku itself
  1. there are 5 stages in this project
  2. I will go over each one here
• The 1st step is to create A number(digit) identifier
  1. The digit classifier is made Using Open CV with better image processing than usually done in MNSIT data base
  2. I have also uploaded a repository on how i made my own model on detecting the correct digit
• the 2nd step is to process the image
  1. The Image goes under 4 processes till the point when logical code recives the array
  2. we find the image Contours(boders)

     

  3. we find the sudoku box

     

  4. we crop it and wrap it into a plane full size image

     

  5. finally we send crop image into 81 parts each containg 1 number of sudoku
• The 3rd step is to make the logical code for finding the solution of sudoku
  1. Backtracking is used for the code
  2. this code is effitient than any brute force methord cus it will 9^81 guesses for it to work for a single cell
• 4th Sending the array to the code we wrote just above
  1. this is send as a flat array and returned in the same manner with the answers
• 5th fianl step
  1. The Answers are printed on black screen dewraped
  2. then it is merged on the orginal