| dc.description.abstract | The application of UAVs in the domains of human disaster emergency has opened new windows to many research efforts that are focusing on the detection of humans using UAV aerial images. Aerial images taken by UAVs play an important role in human detection, since lives depend on the results of the detection system, it is always anticipated that the system should be fast, accurate, and economical for it to fit in for emergency response cases. The analyzed data does not only help in detecting and locating humans in post-disaster management but also leads to the construction of the revival plan hence saving lives. In this paper, we focused on the applicability of the UAV camera-based system in post-disaster management to detect and locate humans, for instance, after the disaster had occurred. We proposed a human detection approach that uses FAST-RCNN (regions with convolutional neural networks) which integrates the concept of machine learning in object detection. With a small dataset created our system resulted in 0.706 Precision and Recall of 0.50 while Eye Spy had 0.004 precision and 0.0052 recall, SPOT with precision of 0.4235 and recall is 0.3697. Our human detection model is easy to construct, design, operate, and train compared to the EYE SPY detection system. | en_US |
