Video Surveillance and Deep Learning Enhancing Security through Suspicious Activity Detection
R. Radhika, A. Muthukumaravel · 2024
This study explores the potential of integrating deep learning technologies with video surveillance to detect suspicious behaviors and thereby enhance security measures. By employing sophisticated artificial intelligence methodologies, specifically Convolutional Neural Networks (CNNs), this study seeks to improve the surveillance systems' ability to detect threats in real-time. Through the implementation of an innovative strategy that merges deep learning models with video analytics, the suggested method enhances the precision of suspicious activity detection. Initial data indicates that security protocols have been enhanced, as there is now greater vigilance towards abnormal behavior and fewer false alarms. The integration of deep learning with video surveillance has the capacity to fundamentally transform security frameworks, potentially safeguarding individuals, and averting attacks in advance. The model has performed accuracy of 0.95, precision of 0.92, recall of 0.94 and processing time of 20ms.