Detection and Quantification of Pavement Defects Using Unmanned Aerial Vehicle Imagery
Özet
Roads have an important role in the development of the country's economy and social structure. Today, with the expansion of transportation networks, routine and proper maintenance of roads has gained importance. Cracks and other types of damage appear on the surfaces of roads due to many factors. The detection and measurement of these damages have an important place in the condition assessment of the roads. The surface defects not only affect the visual appearance of roads but also accelerates the aging of concrete infrastructure, which affects their normal use, resulting in shorter lifespans. It is also a potential threat to safe driving. Early detection of reduced capacity due to this deterioration is, therefore, a priority since timely and accurate detection of damages is of vital importance. Nowadays, various methods that are used for detecting surface damage on roads can be listed as follows: an operator inspecting the damage by using traditional guidelines, microscopic examination of the crack using special tools, taking images by unmanned aerial vehicles and automatically interpreting the surface damage by analyzing collected images. If we compare damage assessment using unmanned aerial vehicles with other methods, we can say that it has many advantages such as less risk of accident, low cost, time savings, and fewer logistics requirements. Due to the expanding and increasing road networks, it is indeed a difficult task to conduct an extensive investigation using traditional methods. Therefore, it is possible to use unmanned aerial vehicles with high-resolution cameras, which have been used frequently recently, for surface damage detection of the investigated road. The detection and quantification of damages can be performed using Deep Learning methods from the collected images. With this method, images that accurately reflect the geometry of the damage can be obtained by UAVs. The proposed thesis study aims to automatically detect potholes and cracks on roads, via drones. The results of this study have the potential of contributing the national and international literature on damage detection on roads with the help of UAVs.
