Enhanced YOLOv7 for identifying state of sliding chair in railway turnout
image: PERFORMANCE OF THE PROPOSED METHOD COMPARED WITH THE BASIC YOLOV7 IN THE DETECTION AND THE STATE IDENTIFICATION OF THE SLIDING CHAIRS
Credit: Cao, Y., et al.
The sliding chair is a crucial component in supporting switch rail conversion in the railway turnout. Poor surface conditions of the sliding chair can lead to incomplete conversions. Due to the harsh environmental erosion and the attack from the wheel vibration, the failure rate of the sliding chairs can reach up to 10%.
To that end, a study published in the KeAi journal High-speed Railway introduced a computer vision-based real-time detection method to reliably check the gap state after the turnout conversion by analyzing monitored gap images.
“For the accurate identification of the state of the sliding chair, we proposed an improved YOLOv7-based method,” explains Yuan Cao, first author of the study. “First, the Convolutional Block Attention Module (CBAM) is integrated into the backbone of YOLOv7 to extract more distinguishable features in the images of the sliding chairs, which helps to classify different states with similarities. To detect the sliding chairs with small-scale exposed regions, we further customized an extra detector for small targets into the YOLOv7 network to improve the accuracy of the detection of such sliding chairs.”
For optimal design of the aspect ratio, the researchers revised the localization loss in the objective function, replacing the original Complete IoU with the Efficient IoU (EIoU), which enhances the precise location of sliding chairs.
"Additionally, K-means++ clustering is applied to the anchor boxes in the training set to handle varying-scale sliding chairs, optimizing the priori of the anchor boxes used in YOLOv7,” says Cao.
The improved YOLOv7 is then use to identify the state of sliding chairs. “When deterioration is detected, timely maintenance strategies can be implemented, preventing incomplete turnout conversion caused by degraded sliding chair conditions,” says Cao.
###
Contact author: Yuan Cao, National Engineering Research Center of Rail Transportation Operation and Control System, Beijing Jiaotong University, Beijing 100044, China, ycao@bjtu.edu.cn.
The publisher KeAi was established by Elsevier and China Science Publishing & Media Ltd to unfold quality research globally. In 2013, our focus shifted to open access publishing. We now proudly publish more than 100 world-class, open access, English language journals, spanning all scientific disciplines. Many of these are titles we publish in partnership with prestigious societies and academic institutions, such as the National Natural Science Foundation of China (NSFC).