Article Details

Title · Authors · Abstract · Keywords · DOI · References

Rail Fastener Condition Detection Based on Improved YOLOv5 Algorithm

DI Jiayu1, HU Luping1, ZANG Kai1, WANG Jiannan2, PAN Guangquan1

(1. School of Mechanical and Electrical Engineering, Xi'an Traffic Engineering University, Xi'an 710300, China; 2.Xi'an Aeronautical Polytechnic Institute, Xi'an 710089, China)

Section：Artificial Intelligence · Vol. 2025, Issue 10 · Pages：34-38 · 2096-4706（2025）10-0034-05 · CLC number：TP391.4 · Document code：A · DOI：10.19850/j.cnki.2096-4706.2025.10.007

Abstract

Aiming at the problems of low detection efficiency, high cost, high manual work intensity and high risk coefficient in traditional rail fastener detection methods, this paper attempts to apply Deep Learning to rail fastener detection. This paper takes rail fasteners as the research object, and combines the CBAM attention mechanism and YOLOv5 model to detect rail fasteners. The experimental results show that the improved YOLOv5 rail fastener detection model can effectively detect rail defects. The detection accuracy is 0.93, the recall rate is 0.86, and the F1 score is 0.92, indicating that the model provides an effective detection method for rail fastener detection.

Keywords

rail fastener; defect classification; YOLOv5; CBAM

References

  1. [1] Lu Zuwen Rail fasteners for railways in China [J]. China Railway, 2005 (7): 25-27+10
  2. [2] Zhang Yuan, Lv Defang, Meng Jianjun, etc Defect detection of rail fasteners based on dual attention and GSSN lightweight [J]. Computer Engineering, 2025, 51 (2): 289-299
  3. [3] Wang Le, Zhou Qian, Fang Yue, etc Method for detecting the fastening status of rail fasteners based on line structured light [J]. Advances in Laser and Optoelectronics, 2021, 58 (16): 399-407
  4. [4] Wang Enhong, Chai Xiaodong, Zhong Qianwen, etc Research on the Detection Algorithm of Lateral Deviation Characteristics of Rail Fasteners [J]. Urban Rail Transit Research, 2021, 24 (5): 142-147
  5. [5] Jiang Xinlan Steel rail fastener detection method using structured region fully convolutional neural network [J]. Computer Science and Exploration, 2020, 14 (11): 1888-1898
  6. [6] Fan Hong, Hou Yun, Li Bailin, etc Adaptive Visual Inspection Algorithm for High Speed Rail Fasteners [J]. Journal of Southwest Jiaotong University, 2020, 55 (4): 896-902
  7. [7] Wang Ling, Zhang Bing, Chen Xiai A computer vision based system for detecting missing nuts in rail fasteners [J]. Computer Engineering and Design, 2011, 32 (12): 4147-4150
  8. [8] Bai Tangbo, Gao Jialin, Yang Jianwei, etc Research on Railway Fastener State Detection Method Based on Improved Mask R-CNN [J]. Railway Architecture, 2021, 61 (6): 140-143
  9. [9] Chen Xiuxin State detection method for rail fasteners based on improved YOLOv5s algorithm [J]. Urban Rail Transit Research, 2024, 27 (S1): 147-151
  10. [10] Zhou Ying Research on Defect Detection Method of Track Fasteners Based on Deep Learning [D]. Chengdu: Sichuan University, 2021

Funding

College Student Innovation Training Program (2024DC07); Research Plan Project of Shaanxi Provincial Department of Education (23JK0531)

About the authors

Di Jiayu (2004-), female, Han ethnicity, from Xi'an, Shaanxi Province, undergraduate student, research direction: detection of abnormal and missing rail fasteners; Corresponding author: Hu Luping (1992-), female, Han ethnicity, from Ankang, Shaanxi, lecturer, master's student, research direction: robot vision.