An adaptive combination algorithm based on deep learning and genetic algorithm for anomalous events detection

Abstract

One of the most widely used human behavior detection methods is anomaly detection, which this article covers. Ensuring a person's safety is a crucial task in every community today due to the ever-increasing actions that can be dangerous, from planned crime to harm from an accident. Classic closed-circuit television is insufficient since a person must always be awake and available to monitor the cameras, which is costly. Also, someone's attention tends to decrease after a certain period of time. Due to these reasons, a surveillance system that is automated and able to detect unusual activities in real-time and give sufferers prompt aid is necessary. It should be noted that the identification process must be completed swiftly and correctly. In this paper, we employ a model based on mixes the machine learning (ML) model, namely genetic algorithms with deep learning (DL). In this study's experimentation, the UCF-Crime dataset was employed. The detection accuracy on the testing sample dataset was equal to 89.90%, while the area under the curve (AUC) was equal to 94.58%. The developed models have demonstrated reliability and the ability to achieve the greatest accuracy when compared to models that have already been designed.