A Profound Review of AI-Driven Crime Detection in CCTV Videos

• Published in: 2024 Sixth International Conference on Computational Intelligence and Communication Technologies (CCICT) pp. 193 - 199
• Main Authors: Pisati, Rithya, Astya, Rani, Chauhan, Priyanka
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19-04-2024
• Subjects: burglary detection; CCTV surveillance system; CNN; Crime detection; Deep learning; Hospitals; LSTM; Public security; ResNet; Reviews; street crime; Surveillance; Surveys; theft detection; Video surveillance; violence; YOLO
• DOI: 10.1109/CCICT62777.2024.00040

Closed Circuit Television (CCTV) surveillance systems are widely utilized in public and private areas in order to increase public safety, such as shopping malls, markets, banks, hospitals, universities, schools, streets, and residential apartments. The accuracy and time to detect and identify the crime and criminal are usually the major goals of security applications. However, because of varying environmental factors, the complexities of human activity, the ambiguous nature of the anomaly, and the absence of appropriate datasets, detecting video anomalies is challenging. This paper surveys the last five years, a comprehensive study of detecting crime from videos, and the recently used dataset. Moreover, a comparison study on different approaches has been performed, which are used for anomalies detection such as crime detection, theft, burglary, violence, street crime, shoplifting and unauthorized access. We have noticed that many deep learning algorithms has outperformed other methods in this field