This project implements an AI handwriting recognition engine trained on the MNIST and EMNIST datasets, developed in Python using deep learning frameworks.
The system employs convolutional neural networks (CNNs) to learn from thousands of handwritten digits and characters, achieving high accuracy in classifying user-provided inputs.
The project explores classic computer vision techniques including image normalization, data augmentation, and CNN-based feature extraction. The trained model is deployed in an interactive interface where users can draw digits or letters directly on a canvas and receive real-time predictions. The project also includes evaluation tools such as confusion matrices and per-class accuracy reports for analyzing model performance.
Github: handwriting identifier
