A Machine Learning-Based OCR System for LaTeX Conversion of Mathematical Expressions

This study focuses on an advanced OCR system developed for the automated conversion of handwritten or printed mathematical expressions into LaTeX format. The proposed solution alleviates students’, teachers’, and researchers’ burdens when cataloging mathematical content, which is often tedious and error-prone, by designing custom algorithms to automate the process with printers and scanners. Using OCR techniques alongside machine learning algorithms, MathOCR identifies mathematical symbols and configuration relationships in the input images. The system then generates LaTeX code representative of the identified structures within proper syntactic boundaries. The experimental results stress the capability of MathOCR to improve productivity and accuracy in mathematical documentation within educational and scholarly environments.