Identifying Images of Chess Pieces Using Deep Learning

Abstract: This research investigates the application of Convolutional Neural Networks (CNNs) for classifying chess pieces from image data. Using a balanced dataset comprising 10,000 labeled images across five chess piece categories—Queen, Rook, Bishop, Knight, and Pawn—we adapt a CNN model architecture initially developed for fruit classification and retrain it on chess piece data. Our approach incorporates data analysis techniques such as brightness distribution, RGB color histograms, and model interpretation through edge-based feature maps. Experimental results confirm that CNNs offer high accuracy and class-level differentiation when applied to structured visual objects like chess pieces.