Suspicious Behavior Detection near Vehicles in University Environment: An Approach using Object Detection and Body Angles
Resumo
Context: With the advancement of smart cities, the University environment demands surveillance camera systems to increase their monitoring capabilities to prevent malicious behaviors without prohibiting people’s circulation. Problem: Universities have large parking lots with many vehicles and face daily security problems, such as robberies and kidnappings, due to the lack of cameras capable of detecting suspicious behavior and alerting security personnel. Solution: Our approach enhances security in the university environment by developing a system capable of recognizing vehicles and individuals, assessing their proximity, and detecting gestures and actions labeled as suspicious behavior while interoperating with camera systems to alert the appropriate security authorities. Information systems theory: This work was conceived based on the General System Theory to interact with pre-existing heterogeneous systems. It relates to the Technological Frames of Reference theory, which involves the perception and interpretation of real-time object detection technology to monitor, alert, and ensure security. Method: Our research method is an experimental, descriptive investigation of collecting quantitative data, and our evaluation is conducted through the proof of concept. Results: Our artifact demonstrated its feasibility by exhibiting good performance, enabling the detection of pre-defined suspicious behaviors near vehicles with a precision of 94,25% and accuracy of 86,99%. Contributions and Impact in the area of information systems: Our contributions are two-fold: From the organization’s perspective, our security system interoperability, generating interoperable alerts; the artifact to detect suspicious behaviors, protecting people in the University environment. Our approach impacts the three pilars from IS area: People, Process and Technology. Additionally, we provide a dataset of security camera videos in university parking lots.
