An Intelligent UAV-Based System for Real- Time Victim Detection and Geolocation in Disaster Scenarios

Abstract

Disasters, both natural and man-made, create serious challenges for search and rescue (SAR) operations, where time is crucial for survival. Traditional response methods often struggle due to hard-to-reach areas, dangerous conditions, and a lack of real-time information. This paper discusses the design, development, and evaluation of an intelligent system based on an Unmanned Aerial Vehicle (UAV) that aims to overcome these challenges. The system offers a complete solution for real-time detection and accurate location of victims in disaster-affected areas. It uses a UAV with a high- resolution camera that streams live video to a ground control station. A state-of-the-art deep learning model, You Only Look Once version 8 (YOLOv8), processes this video feed in real-time to quickly and accurately identify human presence. When detection occurs, the system automatically captures the GPS coordinates, logs a timestamp with an image, and shows the victim's location on an interactive dashboard. This gives first responders actionable intelligence right away, which significantly cuts down search times and improves operational efficiency. The paper also outlines a theoretical framework for expanding the system to a multi-UAV swarm, using Swarm Optimization Algorithms for coordinated search patterns and Delay-Tolerant Networks (DTN) to maintain communication in areas without infrastructure. Performance evaluation, which used the VisDrone aerial imagery dataset, shows the system's high detection accuracy and low latency, confirming its potential as a game-changing tool for modern disaster management