Year: 2024
Type: conference-paper
Abstract: Roads connect distant rural areas to urban areas, offering a quick and easy means of transportation.In Bangladesh, most highways and inner roads of rural and urban areas are damaged at various times throughout the year.Nowadays, road damage detection is an emergency crisis.Manual or traditional road damage detection is a very time-consuming and costly method.In this study, we propose an automated process to identify road potholes and cracks by applying deep learning techniques.YOLOv8, a cutting-edge object detection algorithm, has demonstrated encouraging results in several object detection tasks, but its use in road defect detection has not received much attention.We employed the YOLOv8 model to determine the different cracks, including rutting, potholes, travelling, mending, and shoving.We trained the model and tested it using a dataset of 655 photos of roads with cracks labelled.This study demonstrates YOLOv8's ability to detect road damage.The study's findings can be put into soar the effectiveness of the procedures for inspecting road ware defects.To confirm the efficacy, we also contrasted the YOLOv8 model with other models that are currently in use.The outcomes of the trial verified the efficiency with which the enhanced YOLOv8 performance.There is a 3.85% improvement in AP when compared with others.
Authors: Sumiya Chowdhury Anika, Md Riaj Hossain, Ashraful Tamjeed, Meskat Jahan
Language: English
Subfield: Civil and Structural Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4411098960