This project implements the classic LeNet-5 neural network architecture by Yann LeCun for handwritten digit recognition using the MNIST dataset.
-Model Accuracy: Achieves 99.37% accuracy on the MNIST test set.
-Interactive Demo: Includes a PyGame-based interface that lets you draw digits in real time and see live predictions from the model.
-Train the model by running: python lenet5.py
-Launch the drawing interface: python pygame_painter.py
-Reimplementation of LeNet-5
-A model.h5 file containing the trained model with 99.37% accuracy
-Real-time digit prediction using a trained model
-Simple and interactive UI for testing your own digits