This document describes a handwriting recognition project that uses a convolutional neural network model to identify handwritten digits captured from a webcam. The model was trained on the MNIST dataset and uses TensorFlow and Keras for the deep learning components. It takes input from the webcam using OpenCV, processes the image, and returns the predicted digit as output. Some challenges included data cleaning, selecting model parameters, integrating the dynamic input method from the webcam, and image processing. The project provided learning opportunities around label encoding, activation functions, padding techniques, and combining computer vision and deep learning.