With the rapid speed-up of electrified railway and the construction of high-speed electrified railway, automatic defect detection on the catenary support device is of crucial importance for operation and cost reduction. The paper presents an innovative and effective method based on image processing technologies and deep learning networks for bird-preventing and fastener defect. Aim at the problem of low detection accuracy caused by the small defect target on catenary support device in images, an improved Faster R-CNN network added a top-down-top feature pyramid fusion structure is proposed in this paper. The improved faster R-CNN gets 81.2 mAP which achieves a 5.7% improvement over faster R-CNN in railway dataset and 80.1 mAP which achieves a 5.5% improvement over faster R-CNN in VOC2007 dataset. Furthermore, the method can achieve favorable performance against state-of-the-art methods.
Defect Detection for Bird-Preventing and Fasteners on the Catenary Support Device Using Improved Faster R-CNN
Lect. Notes Electrical Eng.
International Conference on Electrical and Information Technologies for Rail Transportation ; 2019 ; Qingdao, China October 25, 2019 - October 27, 2019
Proceedings of the 4th International Conference on Electrical and Information Technologies for Rail Transportation (EITRT) 2019 ; Chapter : 66 ; 695-704
2020-04-02
10 pages
Article/Chapter (Book)
Electronic Resource
English
British Library Conference Proceedings | 2020
|Detection of Catenary Support Component Defect and Fault
Springer Verlag | 2023
|Detection of Catenary Support System
Springer Verlag | 2016
|