Chess camera AI

Project dedicated to detect the positions of a chessboard with a mobile phone camera or webcam in real time.

Made with:

Development detailed

I think, initially, this project could be solved being divided into three main parts or areas, this are grid recognition, pieces recognition and board visualization.

Grid recognition:

The code of this section can be found at src/grid_recog.ipynb

1. Take the image and convert to greyscale.
2. Convert the image to canny *1
3. Do a hough transform to detect chessboard borders *2
4. Delete the lines with less difference within others, for getting only four borders
5. Find intersections to get the four corners of the board
6. Create a black mask excluiding the quadrangle formed by the corners *3
7. Use cv2.goodFeaturesToTrack() to recognize all inner corners.

Grid recognition Issues:

1. The canny threshold has to be very precise and setted by the user, I think there's probably a better option for this
2. This needs to be done without pieces, before the game starts.
3. With this mask you can't move the board during the game

Images:

Original image	Canny	Hough lines	Board corners	Try of mask :b

Pieces recognition:

The pieces recognition still not in development, will actualize here when it does. Will be done with machine learning, so probably incluiding scikit-learn in future.

Board visualization:

The code of this section can be found at src/board.py

Creates the visualization with Pygame library and svg images of pieces. For moving pieces you call on src/main.py the function board.move(move_notation) The move_notation parameter is a string type variable, Eg: "NG1-F3", who abbreviates from "Knight on G1 to F3"

Board positions is defined by a pandas dataframe:

       A  B  C  D  E  F  G  H
    8  r  n  b  q  k  b  n  r
    7  p  p  p  p  p  p  p  p
    6  .  .  .  .  .  .  .  .
    5  .  .  .  .  .  .  .  .
    4  .  .  .  .  .  .  .  .
    3  .  .  .  .  .  .  .  .
    2  P  P  P  P  P  P  P  P
    1  R  N  B  Q  K  B  N  R

After "pD7-D5" and "NG1-F3":

30/7/23: Created brach "develop" Camera configured

Workflow and Colaborate

This projet is being developed with a git Flow structure, for clean and organizated flow of work.

You're invited to colaborate forking this repo and creating a pull request. Always with a detailed dsecription and clean code.

You can contact me at: [email protected]