YOLO11-WLBS: an efficient model for pavement defect detection.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Road Infrastructure Engineering

Background:

• Pavement defects critically impact traffic safety, durability, and operational efficiency.
• Accurate and real-time identification of these defects is essential for effective management.

Purpose of the Study:

• To develop an enhanced lightweight model for accurate and real-time pavement defect detection.
• To improve upon existing detection frameworks by integrating novel modules.

Main Methods:

• Integration of four improved modules: Wavelet Transform Convolution, Lightweight Adaptive Extraction, Bidirectional Feature Pyramid Network, and Simple Attention into the YOLO11 framework.
• Utilized ablation experiments to validate the contribution of each module.
• Evaluated model performance using precision, recall, F1-score, and mean Average Precision (mAP).

Main Results:

• The YOLO11-WLBS model achieved high performance metrics: precision (0.947), recall (0.895), F1-score (0.895), mAP@0.5 (0.944), and mAP@0.5-0.95 (0.703).
• Demonstrated significant improvements over the baseline YOLO11, including a 6.4% increase in precision and a 15.8% increase in recall.
• Reduced model parameters by 25.5% while enhancing detection accuracy and efficiency.
• Showcased robust performance under challenging conditions (extreme lighting, blurring) and strong cross-dataset generalization.

Conclusions:

• YOLO11-WLBS offers an efficient and robust solution for intelligent pavement defect detection.
• The model's lightweight nature and high accuracy make it suitable for real-time deployment on edge devices.
• Presents practical potential for pavement maintenance and infrastructure monitoring systems.