Computer Vision Based Pothole Road Detection and Recognition
Jiedong Lan, Houyong Wang, Zhengxiong Zhu, Qiang Zhang · 2024
abstract-Pothole road detection and recognition is a type of computer vision that aims to identify roads with potholes through digitized images, but in the pothole detection task, traditional classification algorithms often fail to achieve good results because the features of pothole images are often very variable and complex. To address this problem, this paper establishes a fast, high recognition rate and accurate classification Convolutional Neural Network (CNN) model based on deep computer learning techniques for recognizing the presence or absence of potholes in road images. If the road contains potholes, it is able to label the range of potholes and calculate the size of the pothole area, as well as the proportion of the pothole part to the whole image, and these provide a new solution for road pothole detection.