Life Detection Based on UAVs - Thermal Images in Search and Rescue Operation

Abstract

Disasters arise frequently on a global scale and not only seize human lives but also create difficulties to rescue workers, some factors that affect a victim's chances for rescue is the speed in which they are located and dug out. The rescue team work hard to save people by using methods such as trained dogs and robotics, but the resources and time required to deploy a rescue team remain a major issue that reduces people's chances of survival. In recent years, some systems have been developed to speed up the detection of people in disaster scenarios. Today, UAVs are widely deployed in search operations, as they can seize and control a huge region in a period of time. Moreover, with a base in the sky, UAVs can quickly detect disaster-stricken areas to rescue people. Thus, the accurate detection of individuals in images or videos obtained by drones is a well-suited tool to aid in search and rescue (SAR) missions. Therefore, this research explores the reliability of the yolov4 detection model with the collection of thermal images dataset to address the challenge of bad weather conditions, given that thermal images are not impacted by light or bad weather.