Equation-Solver-CNN

Introduction

Code and dataset for the Handwritten Equation Solver (Computer Vision) project. Uses a CNN model, which has an accuracy of about 97% after being trained on the included dataset.

Dataset source: https://www.kaggle.com/xainano/handwrittenmathsymbols

Technologies

Python version: 3.6
Keras version: 2.4.3
CV2 version: 4.4.0
Numpy version: 1.19.5

Setup

To run this project, download files to a directory. Add Jpeg images of handwritten equations consisting of digits 0-9 and/or the multiplication, subtraction, and addition symbols to directory test_images if necessary. Run the program equation_solver.py using the following command:
$ python3 equation_solver.py