Security Enhancement By Integrating Ai Based Real Time Video Surveillance Solutions Using Deep Svm In University Campuses

Abstract

The inability of traditional surveillance systems to manage real-time anomaly detection and the volume of video data makes it more difficult to maintain safety and security on college campuses. To greatly improve campus security, this research suggests a revolutionary method for real-time video monitoring that combines artificial intelligence with Deep Support Vector Machines (SVM). In contrast to current approaches, the suggested system optimizes accuracy and processing speed by combining deep learning techniques for feature extraction with SVM for classification. Through this integration, the system can reduce the average processing time per frame to 25 milliseconds, reduce false alarms by 30%, and reach a classification accuracy of 92% with precision and recall rates of 89% and 90%, respectively. The system uses SVM for effective activity classification and deep learning for reliable feature extraction when processing live video data from several cameras. The paper emphasizes how this system may be easily included into current infrastructure due to its scalability and versatility. By resolving privacy problems with ethical design considerations, this method not only speeds up danger detection and reaction times but also establishes a standard for real-time surveillance solutions.