Open Access

ARTICLE

Ghost-Attention You Only Look Once (GA-YOLO): Enhancing Small Object Detection for Traffic Monitoring

Xinyue Zhang¹, Yuxuan Zhao², Jeremy S. Smith³, Yuechun Wang⁴, Gabriela Mogos⁵, Ka Lok Man¹, Yutao Yue^6,7,8,9, Young-Ae Jung^10,*

1 School of Advanced Technology, Xi’an Jiaotong-Liverpool University, Suzhou, China
2 School of AI and Advanced Computing, Xi’an Jiaotong-Liverpool University, Suzhou, China
3 Department of Electrical Engineering and Electronics, University of Liverpool, Liverpool, UK
4 Department of Information Science and Technology, Sanda University, Shanghai, China
5 School of Information Engineering, Wenzhou Business College, Wenzhou, China
6 Thrust of Artificial Intelligence and Thrust of Intelligent Transportation, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China
7 Institute of Deep Perception Technology, JITRI, Wuxi, China
8 XJTLU-JITRI Academy of Industrial Technology, Xi’an Jiaotong-Liverpool University, Suzhou, China
9 Department of Mathematical Science Technology, University of Liverpool, Liverpool, UK
10 Department of Artificial Intelligence Software Technology, Sun Moon University, Chung-nam, Republic of Korea

* Corresponding Author: Young-Ae Jung. Email:

Computers, Materials & Continua 2026, 87(2), 75 https://doi.org/10.32604/cmc.2026.075415

Received 31 October 2025; Accepted 13 January 2026; Issue published 12 March 2026

Abstract

Intelligent Transportation Systems (ITS) represent a cornerstone in modern traffic management, leveraging surveillance cameras as primary visual sensors to monitor road conditions. However, the fixed characteristics of public surveillance cameras, coupled with inherent image resolution limitations, pose significant challenges for Small Object Detection (SOD) in traffic surveillance. To address these challenges, this paper proposes Ghost-Attention YOLO (GA-YOLO), a lightweight model derived from YOLOv8 and specifically designed for traffic SOD. To enhance the attention of small targets and critical features, a novel channel-spatial attention mechanism, termed Small-object Extend Attention (SEA), is introduced. In addition, the original C2f module is replaced with a more efficient Cross-Stage Partial (CSP) module, C3k2, to achieve improved feature processing with lower cost. Building upon these designs, a CSP-based Ghost Bottleneck with Attention (CGBA) module is further developed by integrating SEA into C3k2 and is deployed within the FPN–PAN network to strengthen feature extraction and fusion. Compared with similar-scale baseline models YOLOv8n and YOLOv11n, GA-YOLO demonstrates clear performance advantages on the UA-DETRAC dataset. Specifically, GA-YOLO achieves over 3% improvements in precision and mAP@50, along with a 5.6% gain in mAP@50-95, while reducing the parameter count by nearly 10% and computational complexity by 0.5 GFLOPS compared with YOLOv8n. In addition, GA-YOLO outperforms YOLOv11n by 8.6% in precision and 3.2% in mAP@50-95. These results indicate that GA-YOLO effectively balances detection accuracy and computational efficiency. Furthermore, additional evaluations across varying occlusion levels and representative detection models indicate the effectiveness and practicality of GA-YOLO for traffic-oriented SOD tasks.

---

Keywords

---