Publication

Abstract : This paper introduces a novel anomaly detection system aimed at safety and security within a college environment using advanced computer vision and machine learning techniques. The proposed system, AasivU, processes continuous video feeds captured from strategically placed cameras across the campus, with the objective of identifying and responding to behaviors that deviate from the norm, such as smoking in restricted areas and physical altercations. The architecture of the system is designed to handle the complexities of real-time video analysis, incorporating stages such as data preprocessing, feature extraction, and optical flow analysis to enhance the accuracy and robustness of anomaly detection. At the core of the system is a Convolutional Neural Network (CNN), which has been trained on a custom dataset to classify various behaviors into normal and anomalous categories. The system is further equipped with a decision-making module that triggers real-time alerts to campus security personnel when an anomaly is detected, enabling prompt and effective responses to potential security threats. Through extensive testing and evaluation, the system has demonstrated high accuracy in detecting a wide range of anomalies, highlighting its potential as a critical tool for enhancing campus safety. This research not only contributes to the field of computer vision by addressing the challenges of real-time anomaly detection in dynamic environments but also provides a practical solution for improving security protocols in educational institutions.