DefinePK

Abstract

The accurate recognition and computational evaluation of handwritten mathematical expressions present a significant challenge in the domain of intelligent systems and digital education. This complexity is primarily due to the diverse nature of human handwriting and the inherently two-dimensional structure of mathematical notation, which traditional Optical Character Recognition (OCR) systems fail to interpret reliably. To address these limitations, this study introduces a deep learning-based framework employing Convolutional Neural Networks (CNNs) for the classification of individual handwritten symbols. The system is trained on a curated dataset of over 96,000 grayscale images encompassing 13 classes, including numeric digits and basic arithmetic operators. After classification, the identified symbols are reconstructed into complete expressions and evaluated using a programmatic method based on Python’s eval() function. The model achieves a training accuracy of 99.55%, demonstrating its efficacy in symbol recognition. Preprocessing techniques such as grayscale conversion, thresholding, contour extraction, and image normalization ensure consistent and high-quality input. The system’s modular design and low computational overhead make it suitable for real-world deployment, including on embedded and mobile platforms. This work lays a foundation for scalable, efficient, and accurate recognition of handwritten mathematical content, contributing to advancements in educational technologies and human-computer interaction