BansiG/Comparative-study-of-ED-SVM-ANN-and-CNN-for-Face-Recognition-using-SIFT-and-HoG-feature-extractors

This ​ project describes various feature extraction techniques for human face recognition and does a comparative analysis among them. Over past couple of decades, face recognition has been a research area of great interest. There are three techniques primarily studied. The first being SIFT features (Scale Invariant Feature extraction) + Support Vector Machine and ANN. Another approach studied is HOG features(Histogram Of Gradients) with Support vector Machine and Artificial Neural Networks. Most importantly, Convolutional Neural Network is also studied. And Lastly, Euclidean distance based face recognition is also carried out. The purpose of this analysis is to demonstrate the most effective and feasible technique for face recognition in terms of design, implementation and applications. Keywords—Face recognition, Histogram of Gradients , Neural networks, Scale invariant feature transform (SIFT) , Support vector Machines.

Jupyter Notebook