Abnormal Event Detection in Human Activity Using Deep Learning

Abstract

Abnormal event detection, human behavior detection, as well asobject recognition plays a vital role in the creation of a smart CCTV system. These systems make it possibleto detect abnormal events in an environment, abnormal behaviors by humans, and the state of alert in the environment. Machine Vision propertyalong with Machine Learning are used in these systems to detect as well as identify the particular anomalies that arise in the video feed from the CCTV. Frame by frame processing is commonly used and Supervised Learning is the commonly used training method for these systems. However, since the anomalies are of many different kinds and also because it is not feasible to pre- detect and train all types of anomalies, supervised learning is being replaced by unsupervised learning and semi - supervised learning for training the system. This system provides a means of minimising or removing the human workload that has to be put on to manually detect and create an alert on detection of an abnormality in the live feed provided by the CCTV. Also, the system increases the storage efficiency by storing only the abnormal events in original quality and storing the normal scenarios in low quality for archiving. Also, this system provides an extension of creating a distributed abnormality classification system, where only the abnormal events are sent on to different dedicated systems to classify the abnormality.