Dynamic image analysis for abnormal behavior detection

Abstract

Our world is now in such developing state where security is more of a concern rather than privacy for an individual. Nowadays, abnormal behavior detection system plays a very important role in various sectors such as, security, prison, bank etc. Abnormal behavior and its definition is different in many cases. In the vivid sense the definition of abnormal behavior, it is something deviating from the normal or differing from the typical scenario. Moreover, this abnormal behavior detection refers to the problem of finding patterns in data that do not conform to expected behavior. For a particular domain abnormal behavior can be different from the classic definition of abnormality. Detection of abnormal behavior is an important area of research in computer vision and is also driven by a wide application domains, such as dynamic image analysis from a video surveillance. Convolutional neural network made this detection and classification way easier and efficient. In this project we are prompted to detect abnormal or suspicious behavior by an individual person. Our purpose is to detect behavior which is not normal from dynamic images taken from a video surveillance. In this case we are using Convolutional Neural Network (CNN) to detect abnormal behavior. In experiments, our proposed system detected the behavior of individuals in normal scenario successfully with the accuracy of 98%. Moreover, it also detects any deviations from previous data for any new scenario from different dynamic images. Our system can be implemented in advanced security purposes.