Speaker
Description
Road distresses, such as potholes and alligator cracks, are major safety hazards that lead to increased maintenance costs and vehicle damage. Traditional manual inspection methods are time-consuming, subjective, and often inaccurate. This study introduces an automated, real-time road distress detection system that uses deep learning algorithms, specifically Convolutional Neural Networks (CNNs) and YOLOv8, to improve detection accuracy and efficiency across various environmental conditions.
Our approach involves data acquisition, labeling, and training on a diverse dataset of potholes and alligator cracks. The system achieved a detection accuracy of 92% and a processing speed of 95 frames per second (fps), making it ideal for real-time applications. Beyond simple detection, the system also calculates the area and volume of detected potholes, geotags their locations, and visualizes them on an interactive map. All results are stored in a database and can be exported as detailed PDF reports, which supports more efficient road maintenance.
This research has significant implications for intelligent transportation systems and smart city initiatives, enabling proactive maintenance and optimized resource allocation. Future work will focus on improving the system's robustness and deploying it on edge devices, such as mobile platforms, to increase its practical utility.
Keywords: Deep Learning, YOLOv8, Road Distress Detection, Potholes, Alligator Cracks, Real-time Detection, Smart Cities, Automated Road Maintenance
| Final Abstract | f1 |
|---|