DESIGN AND IMPLEMENTATION OF A WEAPON DETECTION SYSTEM FOR SMART CITIES USING DEEP LEARNING

Abstract

Deep learning-based convolutional neural network topologies help build and implement smart city weapons detection systems. The device detects knives and guns in real time using cutting-edge computer vision. Time-sensitive urban security applications can grow inference with low latency using the edge and cloud. A robust data preparation and augmentation method makes the system more tolerant to lighting, occlusion, and crowds. Many labeled datasets are used to train the model for different urban scenarios and camera angles. Transfer learning reduces training time without compromising detection. An intelligent alerting system alerts authorities to high-confidence detections. Accuracy, recall, F1-score, and inference speed are common system performance measurements. Experiments show detection works dependably in difficult real-world situations. This architecture allows smart city systems to install many cameras. Data is processed for privacy and preserved minimally owing to ethics and legislation.