AI-Powered Suspicious Activity Monitoring and Detection System Using CNN Architecture

Abstract

The rapid advancement of artificial intelligence (AI) and deep learning has revolutionized security surveillance systems by enabling real-time detection and monitoring of suspicious activities. This study presents an AI-powered Suspicious Activity Monitoring and Detection System leveraging a Convolutional Neural Network (CNN) architecture for enhanced accuracy and reliability. The proposed system utilizes a hybrid deep learning approach, integrating CNNs with feature extraction techniques to classify and detect anomalous behaviors in real-time surveillance footage. A large-scale dataset comprising diverse human activities, including normal and suspicious actions, was utilized for model training and validation. The CNN model was optimized using transfer learning and hyperparameter tuning, achieving a detection accuracy of 98.3% on benchmark datasets. The system is further integrated with an edge computing framework, ensuring real-time processing and reduced latency in security-critical environments. Performance evaluation metrics such as precision, recall, F1-score, and inference time validate the model’s effectiveness against existing state-of-the-art methods. The results demonstrate that the proposed CNN-based system significantly enhances the efficiency of automated surveillance by reducing false positives and improving threat response mechanisms. This study contributes to the development of robust AI-driven security solutions, fostering safer public and private spaces through intelligent video analytics.