Abstract: Accurate detection of road damage is crucial for ensuring traffic safety and optimizing road maintenance. However, existing methods face significant challenges in dealing with complex backgrounds, multi-scale damage targets, and irregular feature shapes. To address these limitations, this study proposes a road damage detection algorithm based on DRFENet (Damage-Resistant Feature Enhancement Net). DRFENet incorporated a global and local feature enhancement module (GLFEM) to effectively focus on damage regions while suppressing background interference. A multi-layer feature aggregation module (MFA) was designed to enhance the modeling capability for multi-scale damage features. The WiseIoU loss function was employed to optimize the detection performance for irregular damage targets. Experiments conducted on the RDD2022 and UAVROAD public datasets demonstrate that DRFENet achieves mAPs of 87.2% and 86.8%, respectively, significantly outperforming existing mainstream methods while maintaining excellent real-time performance. Further experiments on embedded platforms show DRFENet's high efficiency and robustness in resource-constrained environments.