Smart Pothole Detection Using Deep Learning Based on Dilated Convolution

Abstract

Roads make a huge contribution to the economy and act as a platform for transportation. Potholes in roads are one of the major concerns in transportation infrastructure. A lot of research has proposed using computer vision techniques to automate pothole detection that include a wide range of image processing and object detection algorithms. There is a need to automate the pothole detection process with adequate accuracy and speed and implement the process easily and with low setup cost. In this paper, we have developed efficient deep learning convolution neural networks (CNNs) to detect potholes in real-time with adequate accuracy. To reduce the computational cost and improve the training results, this paper proposes a modified VGG16 (MVGG16) network by removing some convolution layers and using different dilation rates. Moreover, this paper uses the MVGG16 as a backbone network for the Faster R-CNN. In addition, this work compares the performance of YOLOv5 (Large (Yl), Medium (Ym), and Small (Ys)) models with ResNet101 backbone and Faster R-CNN with ResNet50(FPN), VGG16, MobileNetV2, InceptionV3, and MVGG16 backbones. The experimental results show that the Ys model is more applicable for real-time pothole detection because of its speed. In addition, using the MVGG16 network as the backbone of the Faster R-CNN provides better mean precision and shorter inference time than using VGG16, InceptionV3, or MobilNetV2 backbones. The proposed MVGG16 succeeds in balancing the pothole detection accuracy and speed.

Enhanced pothole detection system using YOLOX algorithm

“…Similarly, the corresponding mAP@0.50 values of YOLOv3tiny were obtained from research by Dharneeshkar et al [30]. In Table 3, the corresponding result values of YOLOR-P6, YOLOR-W6, YOLOv5m and YOLOv5s were taken from the results of the research work done by Ahmed [38]. For each parameter, the corresponding bar charts have been plotted to better understand the strengths and weaknesses of our model compared to others.…”

“…It has been said to be a significant improvement over both the YOLO v3 and v4 in terms of both speed and precision. Several research projects were conducted on YOLOv5 [36,38] to test the precision and accuracy of the algorithm. Ahmed [38] proposed to find a best method to detect potholes by evaluating and comparing different parameters of different image recognition algorithms like YOLOv5, YOLOR [39] and Faster R-CNN with five different backbone structures.…”

“…Several research projects were conducted on YOLOv5 [36,38] to test the precision and accuracy of the algorithm. Ahmed [38] proposed to find a best method to detect potholes by evaluating and comparing different parameters of different image recognition algorithms like YOLOv5, YOLOR [39] and Faster R-CNN with five different backbone structures. The results obtained show that Faster R-CNN with ResNet50 backbone has the highest precision, followed by YOLOv5.…”

Dataset of road surface images with seasons for machine learning applications

“…The images are in .jpg format, and the videos are in mp4 format. Due to the complexities of earth surface and remote sensing data, it is necessary to identify road surfaces in various conditions [1] . Thus to overcome the time-dependent and weather variations of illumination in an outdoor environment, the dataset consists of road surface images in the form of speed breaker, uneven road surface, potholes in rains and potholes in summer.…”

“…In transportation infrastructure, one of the main concerns is potholes in roads [1] . Most machine learning techniques for autonomous driving are trained on data collected in certain environments and are not reliable in cross weather conditions [2 , 6] .…”