摘要
为提升车站旅客引导服务效率,提高车站智能化服务水平,深入研究车站旅客人群密度估计,针对铁路车站人群密度差异大、分布不均匀的场景特殊性,构建基于深度神经网络的自适应场景人群密度估计模型,通过引入注意力机制处理模块,不同尺寸人群图像的识别模块和自适应场景权重判断模块,实现了车站不同场景下的人群密度估计。以清河站为试验场景,对现场采集视频图像样本进行训练学习和验证,准确率达到92%以上,验证了方法的可行性和有效性,该研究成果可为铁路车站图像智能化处理提供借鉴和指导。
To improve the efficiency of passenger guidance service and promote the intelligent service in railway stations,this paper studied crowd counting estimation at railway stations.Considering the differences and uneven distribution of passenger density at railway stations,the paper constructed a crowd counting estimation model based on a deep neural network under the self-adaptive scenario.The introduction of the attention mechanism processing module,the recognition module of crowd images of different sizes,and the self-adaptive scenario weight judgment module facilitated the crowd counting estimation at stations under different scenarios.Taking the Qinghe Railway Station as a test scenario,this paper carried out the training,learning,and verification of the video image samples collected on the spot,with the accuracy reaching higher than 92%,which verified the feasibility and effectiveness of the method.The research results can provide a reference and guidance for the intelligent processing of railway station images.
作者
李瑞
李平
王万齐
代明睿
LI Rui;LI Ping;WANG Wanqi;DAI Mingrui(Postgraduate Department,China Academy of Railway Sciences,Beijing 100081,China;Institute of Computing Technology,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China)
出处
《铁道运输与经济》
北大核心
2021年第11期19-26,共8页
Railway Transport and Economy
基金
国家重点研发计划(2020YFF0304104)。
关键词
场景自适应
深度学习
人流密度
注意力机制
铁路车站
Self-Adaptive Scenario
Deep Learning
Crowd Counting
Attention Mechanism
Railway Station
