Monitoring and Mapping of Road Network Conditions based on Multitemporal Ultra-high Spatial Resolution Data

The article discusses the main types of pavement defects characteristic of Russian roads, the factors influencing their occurrence, as well as means and methods of remote monitoring of the road network. To carry out a comprehensive assessment of the state of the road network using data from unmanned aerial vehicles (UAVs) and optimize road repair measures, a research methodology and an integrated approach to compiling large-scale maps and roadway plans providing detailed information about the condition and characteristics of the road infrastructure are presented. The article describes the technological aspects of drone shooting and post-processing, including obtaining orthophotomaps, digital elevation models, multispectral image mosaics, heat maps, multitemporal datasets of different seasons and video data. The study determined the optimal altitude flight levels for the operation of an unmanned aerial vehicle, and also assessed the potential for detecting road defects using digital terrain models and other derived data.The advantages of the methodology for monitoring the road network, which includes shooting and processing data using ultra-high resolution data, are emphasized. A semantic segmentation model based on the YOLOv8 (You Only Look Once) algorithm has also been created to identify classes of road defects from UAV images, which successfully highlights small coating defects and demonstrates low sensitivity to shadows and road markings.