Handwritten Arabic letters recognition

Overview

Machine Learning Python application to recognize Arabic letters using the classifier k-NN and the descriptor LBP.

Dataset

The model was trained on the dataset Arabic Handwritten Characters Dataset. It contains 16800 images of 32x32, each of a signle handwritten arabic letter.

k-nearest neighbors (k-NN)

A supervised classification learning algorithm. It assigns the closest neighboring class to an input.

Local Binary Pattern (LBP)

A visual descriptor used for classification in computer vision. It's mainly used for texture classification. We noticed, upon applying it on the handwritten pictures, an improvement in the accuracy of the k-NN model.

Short demonstration

This application was a part of a school assignment where k-NN and LBP were specifically required. Hence the reason why they were used.

Contributors

Fatiha HAMMA (@05-byhata)