Face Classification- Statistical Modelling, Viola Jones Algorithm and CNN | Pytorch Jan 2020 - Apr 2020 • project 2_data exxtraction :Prepared the dataset (subset of WiderFace Dataset) by parsing annotations and cropping considering the IoU. project 1: Build and compared the performance of Single Gaussian, Mixture of Gaussian, t-Distribution, Mixture of t-Distribution, Factor Analyzer and Mixture of Factor Analyzer based on ROC curve, Classification rates, accuracy (70-90%), F1 score. •project 2: Implemented the Viola Jones algorithm from scratch by extracting Haar features and observed performance by varying the number of weak classifiers in the ensemble. Got 97.60% testing accuracy on the data. •project 3: Further build a feed forward CNN based LeNet5 architecture reaching accuracy of 98.79%.