Handwritten Polynomial Equation Solver up to Degree 3 Using CNN

Abstract

This project presents an easy-to-use system that can recognize and solve hand written polynomial equations using, Convolutional Neural Networks. The process begins with converting an image of a handwritten equation into grayscale, followed by noise removal to clean it up. The individual symbols are then separated using OpenCV, and then a custom-trained CNN model classifies each symbol. To ensure that the model could handle a wide range of handwriting styles, we started with a dataset of over 30,000 digit and math symbol samples and expanded it through data augmentation to more than 100,000 images. This significantly improved the model�s ability to generalize. The CNN, built using Keras, achieved an impressive 98.99% classification accuracy, reliably identifying each character. Once the symbols are recognized, they are pieced back together into a full polynomial equation, which is then solved using symbolic computation techniques. Everything comes together in a user-friendly graphical interface built with PyQt5, where users can upload their handwritten equations, select the equation type, and see the solution interactively. The system supports basic mathematical symbols (+,�, �,=,x,y) and can handle polynomial equations up to the third degree. Whether for students learning algebra or for quick problem-solving, this tool offers a practical and educational solution for interpreting handwritten math.