Handwritten_Digit_Recognition

A non linear deep neural network that recognizes handwritten digits in a 28x28 pixels image. If you wish to test the code with your own handwritten digits, please create your handwritten digits in a 28x28 pixels image and name it "My_Digit.png", and store it in the same folder as "Digit_Recognition_Non_Linear.py" before running the code. "Digit_Recognition.py" is a linear regression model that has no hidden layers. "Digit_Recognition_Non_Linear.py" is a logistic regression model that has two hidden layers to help better predict what a handwritten digit is. The data/Mnist folder contains the famous Mnist dataset which I used to train the model. Libraries required to run the code: PyTorch, Python Image Library.