This code generates FEN code according to the location of chess pieces, using the SGDClassifier machine learning model.

Original file is located at colab

• This code generates FEN code according to the location of chess pieces, using the SGDClassifier machine learning model.
• The method is similar to the famous MNIST problem.
• The training data contains 18,000 chessboards.
• The code then creates 60,000 images from these chessboards, each containing 64 squares, with up to 32 pieces in the game.
• example: FEN code: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR

• cv2
• numpy
• pandas
• glob
• os
• re
• tqdm
• matplotlib

normalizeBoardName(boardName):

this function normalizes the FEN code.

• Introduction of the folder that contains the images.
• Define a function to normalize the FEN code.
• Creation of 60,000 images from the chessboards, where each board contains 64 squares and at most 32 pieces.
• Save the 60,000 images in an Excel file.
• Read the data from the directory.

• Convert categorical features to numerical.
• Split the data into training and testing sets.

• after trainig a SGDClassifier model we can simply predict the fen codes of each chess board

Original file is located at colab

• more info about fen: fen_wiki

• the training data contains 18k chess board ( G_drive)

• the data for trainig the mode(60k-peice)( G_drive )

• the test data (the answers should be checked visually)( G_Drive)

Mnist data can load from google drive: MNIST