基于深度残差神经网络的地铁站台门与列车门间异物自动检测方法研究
孔德龙蒲凡
Research on Automatic Detection Method of Foreign Objects Between Platform Screen Door and Train Door Based on Deep Residual Neural Network
KONG DelongPU Fan
-
作者信息:中南民族大学计算机科学学院， 430074，武汉
-
Affiliation:School of Computer Science, South\|Central University for Nationalities, 430074, Wuhan, China
-
关键词:
-
Key words:
-
DOI:10.16037/j.1007-869x.2021.12.013
-
中图分类号/CLCN:U231.4
-
栏目/Col:研究报告
摘要:
针对地铁站台门与列车门间现有异物检测方法的精度差、误报率高的问题，提出基于深度残差神经网络图像识别原理，利用地铁站台发车指示器图像数据实现站台门与列车门间异物实时检测。首先，搭建基于深度残差神经网络ResNet50模型的自动异物检测系统；然后，采集站台发车指示器视频帧信息建立数据集并完成系统训练；最后，分析自动异物检测系统对验证信息集的处理效果，并将该系统应用于实际地铁车站中。处理效果表明：实际应用验证中最低准确率为98.7%，单张视频帧处理总耗时不超过65 ms，满足地铁实际运营的要求。
Abstracts:
In view of the problems of poor accuracy and high false alarm rate of the existing detection method of foreign object between metro station platform screen door and train door, image recognition principle based on deep residual neural network is proposed, and real-time detection is realized using metro station departure indicator image. First, automatic foreign object detection system based on deep residual neural network ResNet50 model is built; then, platform departure indicator video frame information is collected to establish a data set and to complete system training; finally, the system processing effect of verification information set is analyzed, and the system is applied in actual metro stations. The processing results show that the minimum accuracy rate in actual application verification is 98.7%, and the processing time for single video frame does not exceed 65 ms, which meets the requirements of actual metro operation.
引文 / Ref:
孔德龙，蒲凡.基于深度残差神经网络的地铁站台门与列车门间异物自动检测方法研究[J].城市轨道交通研究，2021，24(12):66.
KONG Delong，PU Fan.Research on automatic detection method of foreign objects between platform screen door and train door based on deep residual neural network[J].Urban Mass Transit，2021，24(12)：66.
KONG Delong，PU Fan.Research on automatic detection method of foreign objects between platform screen door and train door based on deep residual neural network[J].Urban Mass Transit，2021，24(12)：66.
- 上一篇: 高铁物流运输模式及其可行性
- 下一篇: 轨道车辆产品专利管理探析
过刊浏览
资料下载
友情链接