To enhance small object detection in complex road senses, an improved you only look once version 8 (YOLOv8) algorithm is introduced. Firstly, the quadruple down-sampling branch is added to improve the learning ability of the network for small object features. Secondly, a spatial pyramid pooling fast bi-level routing spatial attention (SPPF-BRSA) module is designed to remove irrelevant regions in a query adaptive manner, which effectively reduces the interference of complex background on detection performance. In addition, the C2fDynamic module is used in the neck of YOLOv8 to strengthen the feature expression ability of the model by dynamically selecting the convolution kernel. Finally, the wise intersection over union (Wise-IoU) v3 loss function is used to obtain more accurate detection results by adjusting the gradient gain distribution. The experimental results show that on Huawei SODA10M dataset, the improved algorithm improves precision (P), recall (R), F1 score and mean average precision at IoU threshold of 0.50 (mAP50) by 4.6%, 4.1%, 4.3% and 5.4%, respectively compared with the original algorithm.
| [1] |
ZHU Y, ZHANG Y X. SEP-YOLO: road object detection algorithm based on YOLOv8[J/OL]. Computer applications and software, 2025. http://kns.cnki.net/kcms/detail/31.1260.tp.20240829.1140.002.html. (in Chinese)
|
| [2] |
Sunkara R, Luo T. No more strided convolutions or pooling: a new CNN building block for low-resolution images and small objects. Joint European Conference on Machine Learning and Knowledge Discovery in Databases, September 19–23, 2022, Grenoble, France, 2022, Cham, Springer Nature: 443-459 [C]
|
| [3] |
Sheng B Y, Hou J, Li J X, et al. . Road object detection method for complex road scenes. Computer engineering and applications, 2023, 59(15): 87-96 [J]
|
| [4] |
CHEN Z Z, LIU G W. Research on vehicle detection algorithm based on improved YOLOv8[J/OL]. Journal of Chongqing Technology and Business University (natural science edition), 2025: 1–9. http://kns.cnki.net/kcms/detail/50.1155.N20240717.1427.002.html (in Chinese)
|
| [5] |
Ran X S, Su S J, Chen J H, et al. . Object detection algorithm for complex road scenes based on adaptive feature fusion. Computer engineering and applications, 2023, 59(24): 216-226 [J]
|
| [6] |
Wei C H, Yang R, Liu Z R, et al. . YOLOv8 with bi-level routing attention for road scene object detection. Journal of graphics, 2023, 44(06): 1104-1111 [J]
|
| [7] |
SUN L S, SHI T, DING Y, et al. EBD-YOLO: a lightweight auxiliary driving detection algorithm based on YOLOv5[J/OL]. Optoelectronics letters, 2025. http://kns.cnki.net/kcms/detail/12.1370.tn.20250103.1019.024.html.
|
| [8] |
TONG Z, CHENY, XU Z, et al. Wise-IoU: bounding box regression loss with dynamic focusing mechanism[EB/OL]. (2023-01-24) [2025-03-12]. https://arxiv.org/abs/2301.10051.
|
| [9] |
Redmon J, Divvala S, Girshick R, et al. . You only look once: unified, real-time object detection. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, June 27–30, 2016, Las Vegas, USA, 2016, New York, IEEE: 779-788 [C]
|
| [10] |
Redmon J, Farhadi A. YOLO9000: better, faster, stronger. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, July 21–26, 2017, Honolulu, USA, 2017, New York, IEEE: 7263-7271 [C]
|
| [11] |
REDMON J. YOLOv3: an incremental improvement[EB/OL]. (2018-04-08) [2025-03-12]. https://arxiv.org/abs/1804.02767.
|
| [12] |
Wang C Y, Bochkovskiy A, Liao H M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for realtime object detectors. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 17–24, 2023, Vancouver, Canada, 2023, New York, IEEE: 7464-7475 [C]
|
| [13] |
Zhu L, Wang X J, Ke Z H, et al. . BiFormer: vision transformer with bi-level routing attention. IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 17–24, 2023, Vancouver, Canada, 2023, New York, IEEE: 10323-10333 [C]
|
| [14] |
Chen Y P, Dai X Y, Liu M C, et al. . Dynamic convolution: attention over convolution kernels. IEEE Conference on Computer Vision and Pattern Recognition, June 13–19, 2020, Seattle, USA, 2020, New York, IEEE: 11027-11036 [C]
|
| [15] |
Ma Q W, Liu G H. Surface defect detection method of strip steel based on lightweight YOLOv3. Journal of optoelectronics-laser, 2023, 34(7): 743-751 [J]
|
| [16] |
Zheng Z, Wang P, Ren D, et al. . Enhancing geometric factors in model learning and inference for object detection and instance segmentation. IEEE transactions on cybernetics, 2021, 52(8): 8574-8586 J]
|
| [17] |
HAN J H, LIANG X W, XU H, et al. SODA10M: a large-scale 2D self/semi-supervised object detection dataset for autonomous driving[EB/OL]. (2024-06-21) [2025-03-12]. https://arxiv.org/abs/2106.11118.
|
| [18] |
Du J, Cui S H, Jin M J, et al. . Complex road scene target detection algorithm based on improved YOLOv7. Computer engineering and applications, 2024, 60(01): 96-103 [J]
|
| [19] |
Yang L, Chen Y F, Li H M, et al. . Object detection algorithm for autonomous driving scenes based on improved YOLOv8. Computer engineering and applications, 2025, 61(01): 131-141 [J]
|
RIGHTS & PERMISSIONS
Tianjin University of Technology
PDF
