Comparative Analysis of Traditional and Machine Learning Based Face Recognition Models

Face recognition technology uses methodologies to process images, register them, and compare and verify it with a set of faces it has already registered and labelled. Recent development in visual recognition showcases the immense power of utilising machine learning techniques like deep neural networks, convolutional neural networks to achieve face recognition. Based on a comprehensive literature review it was found that some of the algorithms that provide great performance metrics in a variety of conditions are MTCNN, FaceNet model, Haar cascade, LBPH, hence it was chosen to be the subject of comparison under various conditions in the paper. A machine learning model was constructed which included MTCNN for face detection and FaceNet for face recognition.The Performance of KNN and SVM classifiers has been studied. The second model uses a combination of Haar cascade for face detection and LBPH for face recognition, a traditional face recognition technique. The results highlight that MTCNN, FaceNet and KNN perform exceptionally well. The performance of the other models are also discussed in the results and discussion section. The research not only guides researchers and practitioners in choosing suitable face recognition techniques but also offers ground for future improvements in accuracy and reliability in the field of algorithm developments.