Comparative Study on Rail Damage Recognition Methods Based on Machine Vision
With the rapid expansion of railway networks and increasing operational complexity, intelligent rail damage detection has become crucial for ensuring safety and improving maintenance efficiency. Traditional physical inspection methods (e.g., ultrasonic testing, magnetic flux leakage) are limited in...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-07-01
|
| Series: | Infrastructures |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2412-3811/10/7/171 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849246344889237504 |
|---|---|
| author | Wanlin Gao Riqin Geng Hao Wu |
| author_facet | Wanlin Gao Riqin Geng Hao Wu |
| author_sort | Wanlin Gao |
| collection | DOAJ |
| description | With the rapid expansion of railway networks and increasing operational complexity, intelligent rail damage detection has become crucial for ensuring safety and improving maintenance efficiency. Traditional physical inspection methods (e.g., ultrasonic testing, magnetic flux leakage) are limited in terms of efficiency and environmental adaptability. This study proposes a machine vision-based approach leveraging deep learning to identify four primary types of rail damages: corrugations, spalls, cracks, and scratches. A self-developed acquisition device collected 298 field images from the Chongqing Metro system, which were expanded into 1556 samples through data augmentation techniques (including rotation, translation, shearing, and mirroring). This study systematically evaluated three object detection models—YOLOv8, SSD, and Faster R-CNN—in terms of detection accuracy (<i>mAP</i>), missed detection rate (<i>mAR</i>), and training efficiency. The results indicate that YOLOv8 outperformed the other models, achieving an <i>mAP</i> of 0.79, an <i>mAR</i> of 0.69, and a shortest training time of 0.28 h. To further enhance performance, this study integrated the Multi-Head Self-Attention (MHSA) module into YOLO, creating MHSA-YOLOv8. The optimized model achieved a significant improvement in <i>mAP</i> by 10% (to 0.89), increased <i>mAR</i> by 20%, and reduced training time by 50% (to 0.14 h). These findings demonstrate the effectiveness of MHSA-YOLO for accurate and efficient rail damage detection in complex environments, offering a robust solution for intelligent railway maintenance. |
| format | Article |
| id | doaj-art-ac57aad638a0484f95c47a54fdf8f14e |
| institution | Kabale University |
| issn | 2412-3811 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Infrastructures |
| spelling | doaj-art-ac57aad638a0484f95c47a54fdf8f14e2025-08-20T03:58:31ZengMDPI AGInfrastructures2412-38112025-07-0110717110.3390/infrastructures10070171Comparative Study on Rail Damage Recognition Methods Based on Machine VisionWanlin Gao0Riqin Geng1Hao Wu2School of Information Science and Engineering, Chongqing Jiaotong University, Chongqing 400074, ChinaSchool of Mechatronics and Vehicle Engineering, Chongqing Jiaotong University, Chongqing 400074, ChinaSchool of Civil Engineering, Chongqing Jiaotong University, Chongqing 400074, ChinaWith the rapid expansion of railway networks and increasing operational complexity, intelligent rail damage detection has become crucial for ensuring safety and improving maintenance efficiency. Traditional physical inspection methods (e.g., ultrasonic testing, magnetic flux leakage) are limited in terms of efficiency and environmental adaptability. This study proposes a machine vision-based approach leveraging deep learning to identify four primary types of rail damages: corrugations, spalls, cracks, and scratches. A self-developed acquisition device collected 298 field images from the Chongqing Metro system, which were expanded into 1556 samples through data augmentation techniques (including rotation, translation, shearing, and mirroring). This study systematically evaluated three object detection models—YOLOv8, SSD, and Faster R-CNN—in terms of detection accuracy (<i>mAP</i>), missed detection rate (<i>mAR</i>), and training efficiency. The results indicate that YOLOv8 outperformed the other models, achieving an <i>mAP</i> of 0.79, an <i>mAR</i> of 0.69, and a shortest training time of 0.28 h. To further enhance performance, this study integrated the Multi-Head Self-Attention (MHSA) module into YOLO, creating MHSA-YOLOv8. The optimized model achieved a significant improvement in <i>mAP</i> by 10% (to 0.89), increased <i>mAR</i> by 20%, and reduced training time by 50% (to 0.14 h). These findings demonstrate the effectiveness of MHSA-YOLO for accurate and efficient rail damage detection in complex environments, offering a robust solution for intelligent railway maintenance.https://www.mdpi.com/2412-3811/10/7/171rail damagerecognition methoddeep learningYOLO |
| spellingShingle | Wanlin Gao Riqin Geng Hao Wu Comparative Study on Rail Damage Recognition Methods Based on Machine Vision Infrastructures rail damage recognition method deep learning YOLO |
| title | Comparative Study on Rail Damage Recognition Methods Based on Machine Vision |
| title_full | Comparative Study on Rail Damage Recognition Methods Based on Machine Vision |
| title_fullStr | Comparative Study on Rail Damage Recognition Methods Based on Machine Vision |
| title_full_unstemmed | Comparative Study on Rail Damage Recognition Methods Based on Machine Vision |
| title_short | Comparative Study on Rail Damage Recognition Methods Based on Machine Vision |
| title_sort | comparative study on rail damage recognition methods based on machine vision |
| topic | rail damage recognition method deep learning YOLO |
| url | https://www.mdpi.com/2412-3811/10/7/171 |
| work_keys_str_mv | AT wanlingao comparativestudyonraildamagerecognitionmethodsbasedonmachinevision AT riqingeng comparativestudyonraildamagerecognitionmethodsbasedonmachinevision AT haowu comparativestudyonraildamagerecognitionmethodsbasedonmachinevision |