Math Symbol Classification

AI/ML
Computer Vision
Research

Tech Stack

ANN
LBP
Feature Engineering
Image Classification
HOG

Description

Performed a comprehensive comparative analysis of machine learning classifiers for handwritten mathematical symbol recognition using a dataset of mathematical characters.

Implemented and evaluated multiple feature extraction techniques including Gabor filters, Histogram of Oriented Gradients (HOG), Local Binary Patterns (LBP), and normalized raw pixel representations.

  • The SVM with Gabor filter features delivered optimal performance at 98.9% accuracy
  • Highlighted Gabor's strength in capturing edge orientations and spatial frequency patterns
  • Enhanced expertise in feature engineering and classifier comparison