Railway foreign body intrusion detection based on Improved YOLOv4 network model

Yuan Li; Yan Xu; Chunmiao Wei

doi:10.1117/12.2637512

13 May 2022 Railway foreign body intrusion detection based on Improved YOLOv4 network model

Yuan Li, Yan Xu, Chunmiao Wei

Proceedings Volume 12248, Second International Conference on Sensors and Information Technology (ICSI 2022); 1224818 (2022) https://doi.org/10.1117/12.2637512
Event: 2nd International Conference on Sensors and Information Technology (ICSI 2022), 2022, Nanjing, China

Abstract

To solve the problem of low accuracy of detection and positioning of invading foreign body targets of the scene of railway foreign body intrusion, this paper proposes a railway foreign body intrusion detection method that optimizes the YOLOv4 model. First, the self-built railway foreign body intrusion limit dataset was expanded to solve the model over-fitting problem and improve the generalization ability of the model. Second, replace the backbone feature extraction network in the YOLOv4 model with a lightweight feature extraction structure, which effectively reduces the total number of parameters and deepens the number of network layers. Experiments show that the algorithm has good results and performance in detection accuracy and detection speed, effectively solving the problems of missed detection, misdetection, and low detection accuracy of foreign objects in the current road foreign object intrusion scene.

Citation Download Citation

Yuan Li, Yan Xu, and Chunmiao Wei "Railway foreign body intrusion detection based on Improved YOLOv4 network model", Proc. SPIE 12248, Second International Conference on Sensors and Information Technology (ICSI 2022), 1224818 (13 May 2022); https://doi.org/10.1117/12.2637512

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available