Learning discriminative and generalizable features with multi-branch for person re-identification

Finer-grained local features play a supplementary role in the description of pedestrian global features, and the combination of them has been an essential solution to improve discriminative performances in person re-identification (PReID) tasks. The existing part-based methods mostly extract representational semantic parts according to human visual habits or some prior knowledge and focus on spatial partition strategies but ignore the significant influence of channel information on PReID task. So, we proposed an end-to-end multi-branch network architecture (MCSN) jointing multi-level global fusion features, channel features and spatial features in this paper to better learn more diverse and discriminative pedestrian features. It is worth noting that the effect of multi-level fusion features on the performance of the model is taken into account when extracting global features. In addition, to enhance the stability of model training and the generalization ability of the model, the BNNeck and the joint loss function strategy are applied to all vector representation branches. Extensive comparative evaluations are conducted on three mainstream image-based evaluation protocols, including Market-1501, DukeMTMC-ReID and MSMT17, to validate the advantages of our proposed model, which outperforms previous state-of-the-art in ReID tasks.

NN Search based on Non-uniform Space Partition on Air

It is one of most important challenges in information services based on location to support a huge number of clients to process the NN query for a given query point. A scheme of NN query processing based on wireless data broadcasting is an effective way to overcome the challenge. In this paper, we propose an indexing scheme NSPI (Non-uniform Space Partition Index) for quick NN search over wireless data broadcasting. For implementing the aim, we use a non-uniform spatial partition and provide an index based on the partition for equidistance between indexes on the wireless channel. The index scheme enables the clients to process NN quickly by lessening the time for the clients to meet the index on the channel. To show the effectiveness of the proposed scheme, we compare the access time and tuning time of the clients with existing indexing schemes by simulation studies. The proposed NSPI shows shorter access time than the other existing scheme. Also, NSPI outperforms in the aspect of tuning time than the others.

Borderline Disorder: (De facto) Historical Ethnic Borders and Contemporary Conflict in Africa

We explore the effect of historical ethnic borders on contemporary non-civil conflict in Africa. Exploiting variations across artificial regions (i.e., grids of 50x50km) within an ethnicity's historical homeland, we document that both the intensive and extensive margins of contemporary conflict are concentrated close to historical ethnic borders. Following a theory-based instrumental variable approach, which generates a plausibly exogenous ethno-spatial partition of Africa, we find that grid cells with historical ethnic borders have 27 percentage points higher probability of conflict and 7.9 percentage points higher probability of being the initial location of a conflict. We uncover several key underlying mechanisms: competition for agricultural land, population pressure, cultural similarity and weak property rights.

Borderline Disorder: (De facto) Historical Ethnic Borders and Contemporary Conflict in Africa

We explore the effect of historical ethnic borders on contemporary non-civil conflict in Africa. Exploiting variations across artificial regions (i.e., grids of 50x50km) within an ethnicity's historical homeland, we document that both the intensive and extensive margins of contemporary conflict are concentrated close to historical ethnic borders. Following a theory-based instrumental variable approach, which generates a plausibly exogenous ethno-spatial partition of Africa, we find that grid cells with historical ethnic borders have 27 percentage points higher probability of conflict and 7.9 percentage points higher probability of being the initial location of a conflict. We uncover several key underlying mechanisms: competition for agricultural land, population pressure, cultural similarity and weak property rights.

The spatial competition between containerised rail and sea transport in Eurasia

The competition in space between rail and sea transport is of great significance to the integration of Eurasia. This paper proposes a land and sea transport spatial balance model for container transport, which can extract a partition line on which transport costs by rail and sea are equal given a destination. Four scenarios are discussed to analyse the effects of different factors on the model. Then the model is empirically tested on current rail and sea transport networks to identify the transport competition pattern in Eurasia. The location of destinations, the freight costs, and time costs are the three main factors affecting the model. Among them, time costs are determined by the value of a container and its contents, the interest rate, and by time differences between land and sea transport. The case study shows that Eurasia forms a transport competition pattern with a land area to sea area ratio of about 1:2; this ratio, however, changes to 1:1 when time costs are considered. Further, the land and sea transport balance lines are consistent with the theories of geopolitics, which indicate that the same processes may exist in the spatial pattern of geo-economics and geopolitics in Eurasia. According to the balance lines, we get a spatial partition, dividing Eurasia into the land transport preferred area, the land–sea transport indifference area, and the sea transport preferred area. The paper brings a new perspective to the exploration of geopolitical economic spatial patterns of Eurasia and provides a practical geographic theory as an analytic basis for the implementation of the Belt and Road Initiative.

Partitioning of the pore space based on a non-hierarchical decomposition model

Bio-CAD and in-silico experimentation currently have a growing interest in biomedical applications where scientific data coming from real samples are used to evaluate physical properties. In this sense, analyzing the pore-size distribution is a demanding task to help interpret the characteristics of porous materials by partitioning it into its constituent pores. Pores are defined intuitively as local openings that can be interconnected by narrow apertures called throats that control a non-wetting phase invasion in a physical method. There are several approaches to characterize the pore space in terms of its constituent pores, several of them requiring prior computation of a skeleton. This paper presents a new approach to characterize the pore space, in terms of a pore-size distribution, which does not require the skeleton computation. Throats are identified using a new decomposition model that performs a spatial partition of the object in a non-hierarchical sweep-based way consisting of a set of disjoint boxes. This approach enables the characterization of the pore space in terms of a pore-size distribution. computation. Throats are identified using a new decomposition model that performs a spatial partition of the object in a non-hierarchical sweep-based way consisting of a set of disjoint boxes. This approach enables the characterization of the pore space in terms of a pore-size distribution.

A Wind Tunnel Study on the Correlation between Urban Space Quantification and Pedestrian–Level Ventilation

Correlation research on urban space and pedestrian–level wind (PLW) environments is helpful for improving the wind comfort in complex urban space. It could also be significant for building and urban design. Correlation research is usually carried out in a space with clear urban spatial characteristics, so it is necessary to define the space first. In this paper, a typical urban area in Nanjing, China, is selected as the research object, and a spatial partition method is used to divide the real complex urban space into subspaces. The urban spatial characteristics of such subspaces are quantified using three urban spatial indices: openness (O), area (A), and shape (S). By comparing the quantitative results, 24 (12 pairs) subspaces with prominent urban spatial indices are selected as the correlation research cases. The 24 subspaces also provide a reference for the layout of the measurement points in a wind tunnel experiment. This is a new arrangement for locating the measurement points of a wind tunnel for correlation research. In the experiment, 45 measurement points are located, and the mean wind velocity of four different wind directions at 45 measurement points is experimented. The results clearly show that, when the experimental conditions are the same, the changes of mean wind velocity ratio (UR) of 24 (12pairs) subspaces under the four experimental wind directions are close. The URs of the subspaces are not significantly affected by the wind direction, which is affected more by the subspaces’ spatial characteristics. When making the correlation analysis between mean wind speed ratio and spatial characteristics’ indices, a direct numerical comparison was not able to find a correlation. By comparing the difference values of mean wind speed (△UR) and indices between each pair of subspaces, the correlation between UR and openness of subspaces were found. Limited by spatial partition method, the correlation between UR and the other indices was not obvious.