Real-Time Dynamic Scale-Aware Fusion Detection Network: Take Road Damage Detection as an example

TL;DR

This paper introduces RT-DSAFDet, a real-time, scale-aware UAV-based road damage detection model that significantly improves accuracy and efficiency by adaptively handling damage variability and background interference.

Contribution

The paper proposes a novel multi-scale, adaptive detection model with modules for flexible feature extraction and fusion, enhancing UAV road damage detection performance.

Findings

RT-DSAFDet achieves 54.2% mAP50 on UAV-PDD2023, outperforming YOLOv10-m.

The model reduces parameters to 1.8M and FLOPs to 4.6G, with significant efficiency gains.

On MS COCO2017, RT-DSAFDet matches YOLOv9-t in mAP50-95, with higher mAP50 and fewer resources.

Abstract

Unmanned Aerial Vehicle (UAV)-based Road Damage Detection (RDD) is important for daily maintenance and safety in cities, especially in terms of significantly reducing labor costs. However, current UAV-based RDD research is still faces many challenges. For example, the damage with irregular size and direction, the masking of damage by the background, and the difficulty of distinguishing damage from the background significantly affect the ability of UAV to detect road damage in daily inspection. To solve these problems and improve the performance of UAV in real-time road damage detection, we design and propose three corresponding modules: a feature extraction module that flexibly adapts to shape and background; a module that fuses multiscale perception and adapts to shape and background ; an efficient downsampling module. Based on these modules, we designed a multi-scale, adaptive road damage detection model with the ability to automatically remove background interference, called Dynamic Scale-Aware Fusion Detection Model (RT-DSAFDet). Experimental results on the UAV-PDD2023 public dataset show that our model RT-DSAFDet achieves a mAP50 of 54.2%, which is 11.1% higher than that of YOLOv10-m, an efficient variant of the latest real-time object detection model YOLOv10, while the amount of parameters is reduced to 1.8M and FLOPs to 4.6G, with a decreased by 88% and 93%, respectively. Furthermore, on the large generalized object detection public dataset MS COCO2017 also shows the superiority of our model with mAP50-95 is the same as YOLOv9-t, but with 0.5% higher mAP50, 10% less parameters volume, and 40% less FLOPs.