The implementation of Unmanned Aerial Vehicles in indoor emergency situations improves search and rescue operations due to real-time operational recognition and high-precision navigation with reduced operating periods. Thermal cameras carried by drones are essential for emergency indoor analysis, as they can detect heat signatures in low visibility conditions, such as smoke or darkness, improving search efficiency and safety. This work presents a novel approach that combines the analysis of visual and thermal cameras to enhance emergency response efficiency, providing a robust, autonomous solution for detecting individuals in risk and dangerous situations. This study proposes real-time human detection using visual and thermal images captured by drones in indoor emergency scenarios with the implementation of the specialized object detection algorithm You Only Look Once version 8 due to its high accuracy, speed, and computational efficiency. Different scenarios are proposed to test the reliability of the methodology, analyzing images of humans in several postures and environments. Experimental results demonstrate high detection capabilities and avoid false positives, with accuracies higher than 90%.