Driver Fatigue Detection Method Based on Improved RT-DETR

Ying Li; Yangshan Tang; Xiaoyu Wang

doi:10.63313/JCSFT.9082

Authors

Ying Li School of Automotive and Transportation Engineering, Liaoning University of Technology, Jinzhou, 121000, China Author
Yangshan Tang School of Automotive and Transportation Engineering, Liaoning University of Technology, Jinzhou, 121000, China Author
Xiaoyu Wang School of Automotive and Transportation Engineering, Liaoning University of Technology, Jinzhou, 121000, China Author

DOI:

https://doi.org/10.63313/JCSFT.9082

Keywords:

Fatigue Detection, RT-DETR, Object Detection, NWD

Abstract

To address the problem of poor robustness in driver fatigue detection under complex scenarios such as low illumination and motion blur, this paper proposes an optimization method based on hard sample augmentation. A hard sample dataset containing low illumination, motion blur, and their compound interference is constructed, and a progressive two-stage training strategy is designed to guide the model to adapt to various degradations step by step. Furthermore, a spatial-channel interleaved robustness module (SCIRM) is proposed to enhance the response of eye and mouth features, while the normalized Wasserstein distance (NWD) loss function is adopted to improve the localization accuracy of small targets. The improved RT-DETR model achieves mAP50 of 95.03%, 95.62%, and 94.56% under low illumination, motion blur, and compound interference scenarios, respectively, significantly outperforming the baseline model, while its performance on clear scenes remains almost unchanged. This method effectively enhances the robustness of fatigue detection under complex imaging conditions.

References

[1] Li K, Gong Y, Ren Z. A fatigue driving detection algorithm based on facial multi-feature fusion[J]. IEEE Access, 2020, 8101244-101259.Li K, Gong Y, Ren Z. A fatigue driving detection algorithm based on facial multi-feature fusion[J]. IEEE Access, 2020, 8101244-101259.

[2] LIU S, WANG Y, YU Q. A driver fatigue detection algorithm based on dynamic tracking of small facial targets using YOLOv7: Regular section[J]. IEICE Transactions on Information and Systems, 2023, E106. D(11): 1881-1890.

[3] Li J, Hu R. Application of YOLO-Based face recognition in fatigue driving detection[J]. Advances in Computer, Signals and Systems,2025,9(3).

[4] Cui S, Li H, Liu X, Yang C, Gao H, Yang J, Yang B. Driver fatigue detection based on facial multi-feature fusion computing[J]. Computing, 2025, 107(12): 229-229.

[5] Qu W, Wang Z. Fatigue driving detection based on facial multi-information feature fusion[J]. Journal of Physics: Conference Series, 2024, 2872(1): 012018-012018.

[6] Liu G, Wu K, Lan W, Wu Y. YOLO-FDCL: improved YOLOv8 for driver fatigue detection in complex lighting conditions[J]. Sensors, 2025, 25(15): 4832-4832.

[7] Kaibing Z, Cheng Y, Jie L. Multi-branch and progressive network for low-light image enhancement[J]. IEEE Transactions on Image Processing, 2023.

[8] Ajayi, O. O.. A multimodal systematic review of drivers fatigue detection methodologies, datasets, and models[J]. IEEE Access, 2025, 13: 158266.

[9] Arava M, Sundaram M D. Multi‐fatigue feature selection and fuzzy logic‐based intelligent driver drowsiness detection[J]. IET Image Processing, 2025, 19.1: e70052.

[10] Jie H, Li S, Samuel A. Squeeze-and-Excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019, 42(8): 1-1.

[11] Woo S, Park J, Lee J Y. CBAM: Convolutional block attention module[C]. Proc. of the European Conference on Computer Vision (ECCV). Munich: Springer, 2018: 3-19.

[12] S. Abtahi, M. Omidyeganeh, S. Shirmohammadi, B. Hariri. YawDD: A yawning detection dataset[C]. Proc. ACM Multimedia Systems, Singapore, 2014.