Vandalism Detection Using Surveillance Camera

The rapid rise in security concerns has exposed critical limitations of traditional CCTV systems, which mainly serve as passive recorders without enabling timely intervention. This paper proposes an intelligent vandalism detection system that transforms conventional surveillance into an active, automated security solution. The framework employs computer vision and deep learning techniques, integrating a YOLO-based object detection pipeline with action recognition models to identify behaviors strongly associated with vandalism, including breaking, smashing, and spray painting. Upon detection, the system automatically captures incident evidence, either as an image or short video clip, and delivers real-time notifications to property owners. Unlike conventional monitoring methods, the proposed approach operates effectively on both live video streams and recorded footage, ensuring adaptability across diverse deployment scenarios such as homes, commercial spaces, and public facilities. The architecture is designed to be cost-efficient, scalable, and minimally dependent on manual supervision. By coupling AI-driven visual understanding with multi-channel communication, the system not only strengthens deterrence against malicious acts but also accelerates response time and ensures reliable evidence collection. Evaluation demonstrates that the integration of modern deep learning techniques into surveillance workflows significantly enhances detection accuracy while reducing false alerts. The potential of sophisticated video analytics to close the gap between proactive security intervention and passive surveillance is highlighted by this work.