It is crucial to grasp the status of all the catenary support components along the high-speed railway in real time, and thus conduct efficient fault detection to ensure the safety and stability of the catenary equipment. Therefore, in order to study the positioning and bad state detection of the support suspension device of the high-speed rail catenary, so as to improve the efficiency and automation level of components fault detection, a Yolov5 detection network combining the attention mechanism is proposed, called CA-Yolov5. The attention module is continuously added to the last three convolutional networks of different scales of Yolov5. The mechanism of embedding location information into channel attention can effectively improve the convolutional feature expression ability of the network, and guide the model to learn different channel weights without ignoring the importance of location information. Experiments show that the localization accuracy of 12 categories CSCs (Catenary support components) most performs better when compared with some recent representative networks, thus demonstrating the effectiveness of the proposed method.
Detection Approach Based on an Improved Yolov5 for Catenary Support Components
Lect. Notes Electrical Eng.
International Conference on Electrical and Information Technologies for Rail Transportation ; 2021 October 21, 2021 - October 23, 2021
Proceedings of the 5th International Conference on Electrical Engineering and Information Technologies for Rail Transportation (EITRT) 2021 ; Chapter : 64 ; 560-568
2022-02-22
9 pages
Article/Chapter (Book)
Electronic Resource
English
Detection Approach Based on an Improved Yolov5 for Catenary Support Components
British Library Conference Proceedings | 2022
|Preprocessing of Catenary Support Components’ Images
Springer Verlag | 2023
|Roof Components of Metro Vehicles Detection Algorithm Based on Improved YOLOv5
Springer Verlag | 2024
|STD-Yolov5: a ship-type detection model based on improved Yolov5
Taylor & Francis Verlag | 2024
|