A Novel Technology Stack for Automated Road Quality Assessment Framework using Deep Learning Techniques

Abstract

Road infrastructure plays a pivotal role in supporting societal, economic, and cultural progress. The capacity of a road refers to its ability to handle vehicular volume. Inadequate road capacity and the presence of defects like potholes and cracks result in suboptimal travel conditions and pose significant safety risks for drivers, cyclists, and pedestrians. The regular evaluation of these road quality aspects is essential for effective maintenance. However, current methods for assessing road capacity are time-consuming, subjective, and heavily reliant on manual labor. Moreover, existing deep learning-based approaches for detecting road defects often lack accuracy. To overcome these challenges, a fully automated and accurate system for evaluating road quality is imperative. Thus, the objective of this research work is to propose a novel technology stack for a comprehensive Automated Road Quality Assessment (ARQA) framework designed to assess road quality. The experimental findings demonstrate that the suggested vehicle detection and pothole detection methods work effectively and exhibit enhancements of 18% and 6%, respectively, in comparison to existing approaches.

Keywords

• Vehicle Tracking
• LOS computation
• Pothole Detection
• Crack Segmentation
• YOLO Algorithms
• Segmentation