As the problems of conceptual and representational differences will arise among multi-representations, in- ter-connectivity maintenance among multi-representations exists as a foundational task in building multi-scale...As the problems of conceptual and representational differences will arise among multi-representations, in- ter-connectivity maintenance among multi-representations exists as a foundational task in building multi-scale data model. Since the existing methods are still not satisfactory in practice, the inter-connectivity among multiple representa- tions can be only achieved if the multi-scale model is capable of explicitly inter-relating them and dealing with their differences. So, this paper firstly explores the relation among multiple representations from the same entity, such as multi-semantic, multi-geometry, multi-attributes, hierarchical semantic relations and so on. Based on these, this paper proposes aggregation-based semantic hierarchical matching rules (ASHMR) as the basis of tackling inter-connectivity among multi-representations, and defines the available hierarchical semantic knowledge, namely semantically equal, semantically related and semantically irrelevant. According to different change among multi-representations from dif- ferent types of objects, the applications and techniques of the corresponding hierarchy inter-connectivity matching crite- rion are explored. And taken the road intersections as examples, a case in point is given in details for describing the strategies of inter-connectivity maintenance, showing that this method is feasible to deal with inter-connectivity.展开更多
Point cloud analysis is challenging because of the unordered and irregular data structure of point clouds.To describe geometric information in point clouds,existing methods mainly use convolution,graph,and attention o...Point cloud analysis is challenging because of the unordered and irregular data structure of point clouds.To describe geometric information in point clouds,existing methods mainly use convolution,graph,and attention operations to construct sophisticated local aggregation operators.These operators work well in extracting local information but bring unfavorable inference latency due to high computation complexity.To solve the above problem,this paper presents a novel point-voxel based geometry-adaptive network(PVGANet),which combines multiple representations of point and voxel to describe the point cloud from different granularities and can obtain features of different scales effectively.To extract fine-grained geometric features,we design the position-adaptive pooling operator,which uses point pairs’relative position and feature similarity to weight and aggregate the point features at local areas of point clouds.To extract coarse-grained local features,we design a depth-wise convolution operator,which conducts the depth-wise convolution on voxel grids.With an easy addition,fine-grained geometric and coarse-grained local features can be fused,and we can use the geometry-adaptive fused features to complete the efficient shape analysis of point clouds,such as shape classification and part segmentation.Extensive experiments on ModelNet40,ScanObjectNN,and ShapeNet Part benchmarks demonstrate that our PVGANet achieves competitive performance compared with the related methods.展开更多
基金Project 40471090 supported by the National Natural Science Foundation of China, and 2006-1 by the Open Foundation from Key Lab of Resource Envi-ronment and GIS, Beijing City, China
文摘As the problems of conceptual and representational differences will arise among multi-representations, in- ter-connectivity maintenance among multi-representations exists as a foundational task in building multi-scale data model. Since the existing methods are still not satisfactory in practice, the inter-connectivity among multiple representa- tions can be only achieved if the multi-scale model is capable of explicitly inter-relating them and dealing with their differences. So, this paper firstly explores the relation among multiple representations from the same entity, such as multi-semantic, multi-geometry, multi-attributes, hierarchical semantic relations and so on. Based on these, this paper proposes aggregation-based semantic hierarchical matching rules (ASHMR) as the basis of tackling inter-connectivity among multi-representations, and defines the available hierarchical semantic knowledge, namely semantically equal, semantically related and semantically irrelevant. According to different change among multi-representations from dif- ferent types of objects, the applications and techniques of the corresponding hierarchy inter-connectivity matching crite- rion are explored. And taken the road intersections as examples, a case in point is given in details for describing the strategies of inter-connectivity maintenance, showing that this method is feasible to deal with inter-connectivity.
基金supported by the National Natural Science Foundation of China under Grant Nos.62273034,61973029,and 62076026the Scientific and Technological Innovation Foundation of Foshan under Grant No.BK21BF004.
文摘Point cloud analysis is challenging because of the unordered and irregular data structure of point clouds.To describe geometric information in point clouds,existing methods mainly use convolution,graph,and attention operations to construct sophisticated local aggregation operators.These operators work well in extracting local information but bring unfavorable inference latency due to high computation complexity.To solve the above problem,this paper presents a novel point-voxel based geometry-adaptive network(PVGANet),which combines multiple representations of point and voxel to describe the point cloud from different granularities and can obtain features of different scales effectively.To extract fine-grained geometric features,we design the position-adaptive pooling operator,which uses point pairs’relative position and feature similarity to weight and aggregate the point features at local areas of point clouds.To extract coarse-grained local features,we design a depth-wise convolution operator,which conducts the depth-wise convolution on voxel grids.With an easy addition,fine-grained geometric and coarse-grained local features can be fused,and we can use the geometry-adaptive fused features to complete the efficient shape analysis of point clouds,such as shape classification and part segmentation.Extensive experiments on ModelNet40,ScanObjectNN,and ShapeNet Part benchmarks demonstrate that our PVGANet achieves competitive performance compared with the related methods.
