Security System to Analyze, Recognize and Alert in Real Time using AI-Models

Abstract

The field of security systems has advanced significantly in recent years with the advent of deep learning models. Models that can learn complex patterns and features in data have been used in a variety of security applications, such as facial recognition, intrusion detection, and cyber security. Using authorized databases, this proposal uses security to recognize the suspect's face. Here, this article proposes a different method to analyze the long-term videos using the integration technique with the DNN algorithm to recognize the faces using the DNN, Caffe, and Torch models. This proposal used the concept of extracting only the unique features that are present in the faces, such as moles and cracks in the face, for face recognition. It also introduced the concept of "ear-based recognition," where the ear is also a unique part of the body. This feature will be added during the dataset creation and training itself using the DNN feature extractor, Torch. The dataset created with the help of the seven-angle technique suggests taking the seven different angles of the person to cover all the features, and this proposal also handles the problem of camera tampering using the backward subtraction model.