Collective motion is one of the most fascinating phenomena and mainly caused by the interactions between individuals. Physical-barriers, as the particular facilities which divide the crowd into different lanes, greatl...Collective motion is one of the most fascinating phenomena and mainly caused by the interactions between individuals. Physical-barriers, as the particular facilities which divide the crowd into different lanes, greatly affect the measurement of such interactions. In this paper we propose the physical-barrier detection based collective motion analysis (PDCMA) approach. The main idea is that the interaction between spatially adjacent pedestrians actually does not exist if they are separated by the physical-barrier. Firstly, the physical-barriers are extracted by two-stage clustering. The scene is automatically divided into several motion regions. Secondly, local region collectiveness is calculated to represent the interactions between pedestrians in each region. Finally, extensive evaluations use the three typical methods, i.e., the PDCMA, the Collectiveness, and the average normalized Velocity, to show the efficiency and efficacy of our approach in the scenes with and without physical barriers. Moreover, several escalator scenes are selected as the typical physical-barrier test scenes to demonstrate the performance of our approach. Compared with the current collective motion analysis methods, our approach better adapts to the scenes with physical barriers.展开更多
The feature information of the local graph structure and the nodes may be over-smoothing due to the large number of encodings,which causes the node characterization to converge to one or several values.In other words,...The feature information of the local graph structure and the nodes may be over-smoothing due to the large number of encodings,which causes the node characterization to converge to one or several values.In other words,nodes from different clusters become difficult to distinguish,as two different classes of nodes with closer topological distance are more likely to belong to the same class and vice versa.To alleviate this problem,an over-smoothing algorithm is proposed,and a method of reweighted mechanism is applied to make the tradeoff of the information representation of nodes and neighborhoods more reasonable.By improving several propagation models,including Chebyshev polynomial kernel model and Laplace linear 1st Chebyshev kernel model,a new model named RWGCN based on different propagation kernels was proposed logically.The experiments show that satisfactory results are achieved on the semi-supervised classification task of graph type data.展开更多
基金the National Key Research and Development Program of China (2016YFA0502300)the National Natural Science Foundation of China (Grant No. 61602175)+3 种基金Shanghai Municipal Commission of Economy and Informatization (150809)the Open Research Funding Program of KLGIS (KLGIS2015A05) and BUAA (BUAAVR- 15KF-03)the Fundamental Research Funds for the Central Universities (222201514331)Green Manufacturing System Integration Project of Ministry of Industry and Technology of China (9908000006).
文摘Collective motion is one of the most fascinating phenomena and mainly caused by the interactions between individuals. Physical-barriers, as the particular facilities which divide the crowd into different lanes, greatly affect the measurement of such interactions. In this paper we propose the physical-barrier detection based collective motion analysis (PDCMA) approach. The main idea is that the interaction between spatially adjacent pedestrians actually does not exist if they are separated by the physical-barrier. Firstly, the physical-barriers are extracted by two-stage clustering. The scene is automatically divided into several motion regions. Secondly, local region collectiveness is calculated to represent the interactions between pedestrians in each region. Finally, extensive evaluations use the three typical methods, i.e., the PDCMA, the Collectiveness, and the average normalized Velocity, to show the efficiency and efficacy of our approach in the scenes with and without physical barriers. Moreover, several escalator scenes are selected as the typical physical-barrier test scenes to demonstrate the performance of our approach. Compared with the current collective motion analysis methods, our approach better adapts to the scenes with physical barriers.
文摘The feature information of the local graph structure and the nodes may be over-smoothing due to the large number of encodings,which causes the node characterization to converge to one or several values.In other words,nodes from different clusters become difficult to distinguish,as two different classes of nodes with closer topological distance are more likely to belong to the same class and vice versa.To alleviate this problem,an over-smoothing algorithm is proposed,and a method of reweighted mechanism is applied to make the tradeoff of the information representation of nodes and neighborhoods more reasonable.By improving several propagation models,including Chebyshev polynomial kernel model and Laplace linear 1st Chebyshev kernel model,a new model named RWGCN based on different propagation kernels was proposed logically.The experiments show that satisfactory results are achieved on the semi-supervised classification task of graph type data.
