Article Details

Title · Authors · Abstract · Keywords · DOI · References

Small Object Detection Algorithm Based on SSG-YOLO

CHANG Hong1, NI Huahao1, LI Heshuang1, CAI Zhenyu2

(1.Taizhou Power Supply Branch, State Grid Jiangsu Electric Power Co., Ltd., Taizhou 225300, China; 2.Huzhou Institute of Zhejiang University, Huzhou 313001, China)

Section：Artificial Intelligence · Vol. 2026, Issue 8 · Pages：60-66 · 2096-4706（2026）08-0060-07 · CLC number：TP391.4 · Document code：A · DOI：10.19850/j.cnki.2096-4706.2026.08.011

Abstract

To address the accuracy degradation of YOLOv11 in Small Object Detection, this research proposes an improved detection algorithm named SSG-YOLO. Based on YOLOv11n, the algorithm designs a new multi-scale, high-resolution detection head structure to enhance the perception capability for small objects, introduces the parameter-free attention mechanism SimAM to strengthen the model's spatial response ability, and introduces the SIoU loss function to optimize bounding box regression. Experiments conducted on the VisDrone dataset demonstrate that the model achieves improved Small Object Detection accuracy while maintaining a lightweight architecture. However, the SIoU loss function performs poorly in this task, leading to an overall degradation in model performance, indicating its negative impact on small object localization. The final model achieves a good balance between accuracy and speed. Furthermore, training and validation on a self-constructed dataset (DW) show that the improved model increases mAP50 by 1.7% on the VisDrone dataset while retaining its lightweight characteristics, making it suitable for Small Object Detection and target recognition scenarios such as electrical materials.

Keywords

SSG-YOLO; YOLOv11; detection head; SimAM; Small Object Detection; electrical material recognition

References

  1. [1] REDMON J，DIVVALA S，GIRSHICK R，et al. You Only Look Once: Unified, Real-time Object Detection [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas：IEEE，2016：779-788.
  2. [2] Yu Chengyi, Gao Song, Wang Pengwei, etc Improved YOLOv11 Intelligent Vehicle Dynamic Environment Object Detection Algorithm [J]. Electronic Measurement Technology, 2025, 48 (21): 55-66
  3. [3] Gao Jiajie, Si Yachao UAV image object detection algorithm based on YOLOv8 [J]. Journal of Hebei University of Architecture and Technology, 2024, 42 (4): 241-249
  4. [4] Fang Shaofeng, Zou Ziming, Hu Xiaoyan Identification of Irregular Body Events in the Low Latitude F-Zone Ionospheric Field Based on Deep Learning [J]. Frontiers of Data and Computing Development, 2024, 6 (4): 128-138
  5. [5] Zhao Ziyi, Zhang Daode, Xu Shuangcheng Railway foreign object detection algorithm based on improved YOLOv8 [J]. Journal of Shenyang University of Technology, 2025, 44 (3): 32-38, 46
  6. [6] Tian Changrui, Liao Wei, Xu Zhen Improved stacking medical device detection algorithm for YOLOv5s [J]. Chinese Journal of Medical Physics, 2025, 42 (2): 220-226
  7. [7] GAO J B，GUO Y X，GAO G W. Sim-ConvFormer: A Lightweight Fault Diagnosis Framework Incorporating SimAM and External Attention [J/OL].The Journal of Supercomputing，2025，81（4）：603（2025-03-11）.https://doi.org/10.1007/s11227-025-07075-3.
  8. [8] DU D，ZHU P，WEN L，et al. VisDrone-DET2019: The Vision Meets Drone Object Detection in Image Challenge Results [C]//Proceedings of the IEEE/CVF international conference on computer vision workshops. Seoul：IEEE，2019：227-235.
  9. [9] Liu Sicheng, Li Jiachen, Deng Hao, etc Improved Small Object Detection Algorithm Based on YOLOv5 [J]. Ordnance Automation, 2022, 41 (12): 78-82, 94
  10. [10] MO H，WU J W，XIA H，et al. A Lightweight, Efficient, Adaptive Design of YOLOv5 for Enhanced SAR Ship Detection [J].Remote Sensing Letters，2025，16（5）：549-559.
  11. [11] GANESH B R，GEETHA T S，NIRUBAN R. Adaptive YOLOv6 With Spatial Transformer Networks for Accurate Object Detection and Multi-Angle Classification in Remote Sensing Images [J].Expert Systems With Applications，2025，282：127796.
  12. [12] CHEN W，LU J Y，PEI T T，et al.YOLOv8-AFA: A Photovoltaic Module Fault Detection Method based on Multi-scale Feature Fusion [J].Energy Sources, Part A: Recovery, Utilization, and Environmental Effects，2025，47（1）：657-676.
  13. [13] Fang Xiaojie, Yan Liqiang, Zhang Fuhao, etc Research on tomato leaf disease identification based on improved YOLOX-Nano [J]. Anhui Agricultural Science, 2025, 53 (6): 238-242
  14. [14] TAN M，PANG R，LE Q V. Efficientdet: Scalable and Efficient Object Detection [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle：IEEE，2020：10781-10790.
  15. [15] Du Changhao, Zhang Zhi Research on Improved FCOS Algorithm for Vehicle Detection Method [J]. Computer Applications and Software, 2024, 41 (6): 257-262281

Funding

Key R&D Program Project High end Equipment Special Project (2025ZD201)

About the authors

Chang Hong (1979-), male, Han ethnicity, born in Taizhou, Jiangsu, is a senior economist with a bachelor's degree. His research interests include green supply chain, smart warehousing, and image processing; Ni Huahao (1975-), male, Han ethnicity, from Qidong, Jiangsu, senior engineer, bachelor's degree, research direction: green supply chain, data analysis; Li Heshuang (1990-), female, Han ethnicity, born in Jingjiang, Jiangsu, is a senior engineer with a bachelor's degree. Her research interests include green supply chain and smart warehousing; Cai Zhenyu (1994-), male, Han nationality, from Zhuzhou, Hunan, assistant researcher, master's degree, research direction: deep learning, image processing.