With more and more people choose to travel by EMU train, several passengers in railway station with some purposes, for example curiosity or free-tickets, invade the railway line at the end of platform occasionally. This behavior threatens the normal railway operation directly. Therefore, to deal with the difficulties caused by illegal end-intrusion occurring at platform, this paper introduced a method based deep learning for detecting end-intrusion for non-staff using surveillance video. The proposed framework utilizes YOLO v4 to segregate non-staff from the background and Discriminative Correlation Filter (DCF) approach to track the identified non-staff related to end-intrusion with the help of bounding boxes and assigned IDs. Firstly, this paper contributes a high-quality Staff classification dataset from railway stations, named Railway Station Staff Dataset (RSSD, a total of 35262 images including 8 categories). It raises a brand-new practical challenge of image data of railway station. Then, the RSSD were used to finetune pre-trained model. Finally, From the experimental result, it is observed that the YOLO v4 with DCF techniques demonstrated good performance to monitor the end-intrusion at platform. Furtherly, the economic and feasibility assessment are finished. It indicates that the proposed method can be utilized in practical economically and real-time.
Non-staff Detection for End-Intrusion in Railway Station Platform Based on YOLO and DCF Techniques
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 : 20 ; 184-192
2022-02-19
9 pages
Article/Chapter (Book)
Electronic Resource
English
Non-staff Detection for End-Intrusion in Railway Station Platform Based on YOLO and DCF Techniques
British Library Conference Proceedings | 2022
|Fast Obstacle Detection Platform Based on Yolo
British Library Conference Proceedings | 2022
|Fast Obstacle Detection Platform Based on Yolo
Springer Verlag | 2022
|Fast Obstacle Detection Platform Based on Yolo
TIBKAT | 2022
|