Monitoring Windblown Sand along Railway Infrastructures Using Automated Processing of Drone Imagery

Automated image processing technologies are being implemented increasingly for infrastructure monitoring applications thanks to their high level of accuracy, versatility and efficiency. Therefore, they can be employed for management, repair, and maintenance of transport meta-structures to reduce costs, enhance accuracy, and accelerate task completion. Moreover, the novel technology of unmanned aerial vehicles (UAVs) had a considerable effect on the improvement of different industries. This study addresses the influx of sand to the rail surface in desert areas. This problems is now analyzed through the conventional method (i.e. field observation), which is the simplest possible technique requiring ongoing inspection and large amounts of time and resources. Hence, the use of novel technologies of image processing and artificial intelligence as well as aerial images helped take a major step in the automated analysis of this problem. In rail networks across desert areas, the influx and accumulation of sand block railways, slow down and stop trains, and sometimes cause accidents. Thus, it is essential to adopt an automated system for the timely detection and resolution of this problems in order to prevent additional costs or further accidents. UAV-based imaging, smart image processing based on convolutional neural networks, and deep learning are employed in this study to design a comprehensive system for monitoring the problems. This system can greatly decrease the costs of monitoring the safety of rail networks and significantly improve the accuracy and efficiency of repair and maintenance systems. The proposed system can also provide the centralized ongoing monitoring of Iran’s transport network.