Exploring Anomaly Detection Techniques for Crime Detection

Crime anomaly detection is critical for proactive law enforcement and public safety measures. This paper emphasizes identifying and detecting anomalous events harboring criminal intent using the applications of deep learning techniques. Leveraging current research about neural networks, the study explores multiple approaches using pre-trained neural network architectures, including VGG19, DenseNet121, ResNet50, and MobileNetV2 to categorize criminal behavior into multiple classes such as Arrest, Assault, Explosion, Fighting, Road Accident, Robbery, Shooting, Shoplifting, Stealing, Vandalism, Abuse, Arson, Burglary, and Normal Events. The research systematically evaluates the effectiveness of each model by assessing its performance with diverse metrics in detecting anomalies within the UCF Crime Dataset. The findings converge to DenseNet121 model garnering the most accuracy at 82.91%. The proposed methodology provides a foundation for future research in refining crime prediction systems, contributing to advancements in law enforcement technologies.