Smart Surveillance System

The rapid growth of unmanned aerial vehicles (UAVs) has introduced new security challenges, particularly in restricted and high-security areas. Unauthorized drone intrusions pose threats such as espionage, smuggling, and disruption of critical infrastructure. This project proposes a smart surveillance system that detects and classifies unauthorized drones using deep learning. The system integrates the YOLOv5 object detection framework enhanced through transfer learning to distinguish drones from visually similar airborne objects, particularly birds. A curated drone–bird dataset was prepared for training, validation, and testing. The developed system achieves high real-time accuracy, robustness, and efficient aerial threat identification, making it suitable for deployment in sensitive environments. Experimental results demonstrate strong precision, recall, and mAP performance, confirming the effectiveness of the approach in improving situational awareness and countering unauthorized drone activities.