International Journal of Science, Engineering and TechnologyAuthors: Mr. Kunal Kanchankar, Dr. Ravindra Kale
Abstract: The quick growth of the cities and smart cities made the need for surveillance systems that can automatically spot incidents and make the public safer even greater. People are in charge of watching video feeds in the traditional surveillance systems, which can be slow it and make mistakes when there is a lot of footage to watch. Recently advancements in (AI) artificial intelligence, Computer Vision, and deep learning have facilitated automated-systems in the analysis of real-time video streams, enhancing an accuracy of abnormal activity detection. This review paper looks at current research on the AI-based surveillance systems that are meant to find incidents in cities. It talks about the popular methods like (CNN), (RNN), and real-time object detection models like YOLO. The study also reviews available surveillance datasets, evaluation metrics, and implementation approaches used in previous work. A comparison of existing methods highlights their strengths and limitations. The paper also emphasizes the need for integrated multi-incident detection systems capable of improving safety in modern smart cities.
DOI: http://doi.org/10.5281/zenodo.20175429
