Project Description:
This project implements a Viola-Jones facial detection cascade which is trained to detect or reject images based on whether they contain a face. It uses Haar-like features over 24x24 greyscale pixel windows and builds a sequence of strong classifiers with the AdaBoost methodology. It has three phases, whereby it generates more than 100,000 possible features, picks the most effective features in training, and then classifies images in the inference stage. This project emphasizes software design principles as well as data-flow architecture.