Work

Classification of different size of potholes based on surface area using convolutional neural network

Year: 2024

Type: article

Abstract: Roads are essential for transportation, connecting people, moving goods, and supporting business growth. In countries like Pakistan and other South Asian regions, heavy rains can cause significant road damage. Asphalt and concrete are commonly used in road construction, but potholes—bowl-shaped depressions in the road surface—are a major concern. These form when water expands and freezes, exacerbating cracks in the pavement. Potholes worsen as water erodes loose surface particles, and their size can lead to vehicle damage and severe accidents, particularly on high-speed highways. One of the processes of pothole formation is shown in Fig. 1 . Potholes can also (less frequently) result from a variety of non-structural causes, such as diesel (or other chemicals) spillages, mechanical damage to road surfaces from vehicle rims, accidents, and fires, damage caused by falling rocks in cuttings, animal hooves on the road surface in hot weather, Poor road design over specific subgrades (expanding, collapsing, dispersing), lack of bond between the surfacing and WBM base, and insufficient bitumen [ 1 ]. Pothole detection is essential to road maintenance and repair because of its negative impact on road quality and vehicle safety [ 2 ]. Potholes developed due to poor materials, a poor design that causes surface water to accumulate, and ice formation in cracks. Potholes affect the life of people and property every year [ 3 ]. Since 2011, for five years, drivers have spent more than $3 billion on cars to fix damage caused by potholes. The approximate cost per driver comes up to $300 [ 4 ]. According to an India Economic Times story from 2018 [ 5 ], the Supreme Court found 3597 casualties related to highway potholes. Potholes were India’s leading causes of accidents in 2019, accounting for nearly 4,770 incidents. World Health Organization (WHO) states that in 2010 1.25 million people died from road accidents [ 6 ].

Source: Discover Applied Sciences

DOI: 10.1007/s42452-024-06207-3