Face Recognition-Based Surveillance System

In today’s world, technology plays a pivotal role in providing real-time surveillance. Most modern-day surveillance systems are either expensive or not as fully effective when exposed to low illumination or occlusions. The present work aims to develop a comprehensive facial recognition system that is real-time, efficient and reliable. In this work, we use video frames, which are captured from low-powered devices. The algorithm first detects the facial region in each frame by computing the facial embeddings. These embeddings are used as input to machine learning algorithms for performing face recognition. We compare various machine learning algorithms such as Support Vector Machine, k-Nearest Neighbor and Multilayer Perceptron to evaluate their performance for face recognition tasks. Experimental results show that the presented work can be effectively used for domestic surveillance devices, one that is cheap and effective.