Education, Science, Technology, Innovation and Life
Open Access
Sign In
  1. Home
  2. Journals
  3. Journal of Image Processing Theory and Applications
  4. Vol 9, Issue 1, 2026

A Multi-Class Traffic Sign Detection Algorithm Based on an Improved YOLOv13

Download as PDF

DOI: 10.23977/jipta.2026.090105 | Downloads: 7 | Views: 133

Author(s)

Guolin Cong 1, Su Xu 1, Chong Zheng 1

Affiliation(s)

1 School of Electronic Engineering, Jiangsu Ocean University, Lianyungang, Jiangsu, 222000, China

Corresponding Author

Su Xu

ABSTRACT

To address the problems of missed detection of long-distance small traffic signs, insufficient robustness in complex road environments, and limited deployment on vehicle-mounted embedded platforms, a lightweight traffic sign detection algorithm, termed TS-YOLOv13, is proposed based on YOLOv13n. First, the original backbone network is replaced with ShuffleNetV2 to reduce model parameters and memory access overhead while enhancing shallow feature extraction. Second, a lightweight Slim Neck based on GSConv is introduced to improve the efficiency of multi-scale feature fusion. Third, the detection head is reconstructed using the Mobile Inverted Bottleneck Convolution (MBConv), which further reduces computational cost while maintaining detection performance. Finally, the Inner-CIoU loss function is employed to optimize bounding box regression, thereby improving the localization accuracy of hard samples and small traffic signs. To evaluate the effectiveness of the proposed method, ablation and comparative experiments are conducted on the TT100K dataset, while its generalization capability is further validated on the CCTSDB, GTSDB, and LISA datasets. Experimental results demonstrate that, compared with the original YOLOv13n, TS-YOLOv13 achieves a precision of 70.5%, a recall of 57.0%, and an [email protected] of 63.8%. Meanwhile, the number of parameters is reduced from 2.45M to 2.07M, the computational complexity is decreased from 6.2GFLOPs to 5.1GFLOPs, and the model size is reduced from 5.4MB to 4.6MB. Furthermore, compared with several mainstream lightweight object detection algorithms, TS-YOLOv13 achieves a better trade-off among detection accuracy, model size, and inference efficiency, while maintaining good generalization performance across multiple public datasets. The proposed algorithm improves detection accuracy while achieving model lightweighting, making it suitable for real-time traffic sign detection on vehicle-mounted embedded platforms.

KEYWORDS

Traffic Sign Detection; YOLOv13n; Lightweight; ShuffleNetV2; GSConv; MBConv; Inner-CIoU

CITE THIS PAPER

Guolin Cong, Su Xu, Chong Zheng. A Multi-Class Traffic Sign Detection Algorithm Based on an Improved YOLOv13. Journal of Image Processing Theory and Applications (2026) Vol. 9, No.1, 41-57. DOI: http://dx.doi.org/10.23977/jipta.2026.090105.

REFERENCES

[1] Liu L P. Improved canny edge detection algorithm matches traffic signs[J]. Proceedings of SPIE, 2016, 10033.
[2] Setiyono B, Wijaya R N R, Sulistyaningrum D R, et al. Vehicle counting and classification in varying traffic conditions using Faster R-CNN[J]. Journal of Physics: Conference Series, 2025, 2942(1): 012025.
[3] Omodaratan B, Jamali A, Wiley T, et al. Advances in You Only Look Once (YOLO) algorithms for lane and object detection in autonomous vehicles[J]. Engineering Applications of Artificial Intelligence, 2026, 168: 113893.
[4] Han J, Wang C, Du J. HMS-YOLO: Multi-scale traffic sign detection for complex scenarios[J]. Digital Signal Processing, 2026, 183: 106370.
[5] Ma S, Zhao N, Li J, et al. MFSF-YOLO: An improved YOLO11 model for traffic sign detection and recognition[J]. Engineering Letters, 2026, 34(6).
[6] Dong W Y, Chen Y X, Zou G Y. Autotrinet YOLO triple attention framework for robust traffic sign detection[J]. Scientific Reports, 2025, 15(1): 42247.
[7] Xu J, Du Y, Yi Y, et al. An improved lightweight algorithm for traffic sign detection[J]. Scientific Reports, 2025, 15(1): 33554.
[8] Guo S, Zhao N, Ouyang X, et al. RBL-YOLOv8: A lightweight multi-scale detection and recognition method for traffic signs[J]. Engineering Letters, 2024, 32(11).
[9] Zhang J, Yi Y, Wang Z, et al. Learning multi-layer interactive residual feature fusion network for real-time traffic sign detection with stage routing attention[J]. Journal of Real-Time Image Processing, 2024, 21(5): 176.
[10] Zhang X, Tian Y. Improved YOLOv5s traffic sign detection[J]. Engineering Letters, 2023, 31(4).
[11] Han T, Sun L, Dong Q. An improved YOLO model for traffic signs small target image detection[J]. Applied Sciences, 2023, 13(15).
[12] Li D, Hu Y, Cao W, et al. A lightweight fabric defect detection method based on improved ShuffleNetV2[J]. Journal of Real-Time Image Processing, 2026, 23(1): 44.
[13] Huang L, Lai X, Lin P, et al. Lightweight vehicle damage detection using GSConv-based Slim-Neck and bi-level routing attention[J]. World Electric Vehicle Journal, 2026, 17(6): 290.

Subscription
E-Mail Alert
Most popular papers
Call for Special Issue
Publication Ethics & OA Statement
Plagiarism Policy
Contact Us
Downloads: 3198
Visits: 267794

Sponsors, Associates, and Links

All published work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright © 2016 - 2031 Clausius Scientific Press Inc. All Rights Reserved.