Enhanced Human Detection in Disaster Zones Using UAVs and Deep Learning Algorithms

Monitoring and rescuing humans and animals during disasters is essential for reducing their impact on both the environment and human populations. This paper proposes an efficient and reliable system for human detection in catastrophe scenarios using Unmanned Aerial Vehicles (UAVs) equipped with deep learning techniques. In times of natural disasters, such as earthquakes, floods, or wildfires, conventional methods of search and rescue operations often face challenges due to the complexity of the affected environment, limited accessibility, and time sensitivity. The integration of UAVs and deep learning algorithms presents a promising solution to enhance the speed, accuracy, and safety of locating and rescuing survivors in such scenarios. Traditional search and rescue efforts often encounter significant challenges in accessing remote or hazardous areas, leading to delays in locating survivors and providing timely assistance. Our proposed system combines the YOLOv8 object detection model with the Deep SORT object tracking algorithm for human detection. By integrating these advanced techniques, our approach enables accurate and efficient detection and tracking of human subjects in complex environments. The fusion of YOLOv8 and Deep SORT results in a reliable system capable of detecting and tracking moving people in low-quality videos. The combined algorithm, referred to as the Y-DS model, achieved an ultimate accuracy of 87.9<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> and a speed of 55.8 frames per second (FPS), demonstrating effectiveness in real-world scenarios.