Title: Deep learning-based object detection between train and rail transit platform door
Authors: Fen Cheng; Hao Cai
Addresses: School of Machinery and Electronics, Wuhan Railway Vocational and Technical College, Wuhan, Hubei, China ' Network Information Centre, Wuhan Railway Vocational College of Technology, Wuhan, Hubei, China
Abstract: When the train is running, if foreign objects are stuck between the platform tracks, it will often cause huge safety accidents. Therefore, based on deep learning, this paper designs a wide-gap outdoor platform foreign object detection system to detect foreign objects between rail transit platform doors and trains. The main design idea of the system is to explore the real environment of the subway, create a virtual experimental environment, design the hardware equipment of the detection system, and obtain video images captured under various weather and lighting conditions through image analysis and processing. The research results show that the system designed in this paper can well detect the influence of pedestrians getting on and off and intermediate objects when the train stops at the platform, and the detection accuracy of large objects can reach 100%.
Keywords: deep learning technology; machine vision; train rail transit platform; target detection.
DOI: 10.1504/IJGUC.2022.126175
International Journal of Grid and Utility Computing, 2022 Vol.13 No.5, pp.526 - 537
Received: 21 Jul 2021
Accepted: 12 Sep 2021
Published online: 14 Oct 2022 *