As a kind of flexible three-dimensional geometric data, point clouds can accomplish many challenging tasks so long as the rich information in the geometric topology architecture can be deeply analyzed. On account of t...As a kind of flexible three-dimensional geometric data, point clouds can accomplish many challenging tasks so long as the rich information in the geometric topology architecture can be deeply analyzed. On account of that point cloud data is sparse, disordered and rotation-invariant, the success of convolutional neural network in 2 D image cannot be directly reproduced on point cloud. In this paper, we propose WECNN, namely, Weight-Edge Convolution Neural Network, which has an excellent ability to utilize local structural features. As the core of WECNN, a novel convolution operator called WEConv tries to capture structural features by constructing a fixed number of directed graphs and extracting the edge information of the graph to further analyze the local regions of point cloud. Moreover, a weight function is designed for different tasks to assign weights to the edges, so that feature extractions on the edges can be more fine-grained and robust. WECNN gets overall accuracy of 93.8% and mean class accuracy of 91.6% on Model Net40 dataset. At the same time, it gets a mean Io U of 85.5% on Shape Net Part dataset. Results of extensive experiments show that our WECNN outperforms other classification and segmentation approaches on challenging benchmarks.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
Fine-grained visual parsing, including fine-grained part segmentation and fine-grained object recognition, has attracted considerable critical attention due to its importance in many real-world applications, e.g., agr...Fine-grained visual parsing, including fine-grained part segmentation and fine-grained object recognition, has attracted considerable critical attention due to its importance in many real-world applications, e.g., agriculture, remote sensing, and space technologies. Predominant research efforts tackle these fine-grained sub-tasks following different paradigms, while the inherent relations between these tasks are neglected. Moreover, given most of the research remains fragmented, we conduct an in-depth study of the advanced work from a new perspective of learning the part relationship. In this perspective, we first consolidate recent research and benchmark syntheses with new taxonomies. Based on this consolidation, we revisit the universal challenges in fine-grained part segmentation and recognition tasks and propose new solutions by part relationship learning for these important challenges. Furthermore, we conclude several promising lines of research in fine-grained visual parsing for future research.展开更多
基金Supported by the National Natural Science Foundation of China (61772328)。
文摘As a kind of flexible three-dimensional geometric data, point clouds can accomplish many challenging tasks so long as the rich information in the geometric topology architecture can be deeply analyzed. On account of that point cloud data is sparse, disordered and rotation-invariant, the success of convolutional neural network in 2 D image cannot be directly reproduced on point cloud. In this paper, we propose WECNN, namely, Weight-Edge Convolution Neural Network, which has an excellent ability to utilize local structural features. As the core of WECNN, a novel convolution operator called WEConv tries to capture structural features by constructing a fixed number of directed graphs and extracting the edge information of the graph to further analyze the local regions of point cloud. Moreover, a weight function is designed for different tasks to assign weights to the edges, so that feature extractions on the edges can be more fine-grained and robust. WECNN gets overall accuracy of 93.8% and mean class accuracy of 91.6% on Model Net40 dataset. At the same time, it gets a mean Io U of 85.5% on Shape Net Part dataset. Results of extensive experiments show that our WECNN outperforms other classification and segmentation approaches on challenging benchmarks.
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.
基金supported in part by National Natural Science Foundation of China(Nos.62132002,61825101 and 62202010)the Key-Area Research and Development Program of Guangdong Province,China(No.2021B0101400002)the China Postdoctoral Science Foundation(No.2022M710212).
文摘Fine-grained visual parsing, including fine-grained part segmentation and fine-grained object recognition, has attracted considerable critical attention due to its importance in many real-world applications, e.g., agriculture, remote sensing, and space technologies. Predominant research efforts tackle these fine-grained sub-tasks following different paradigms, while the inherent relations between these tasks are neglected. Moreover, given most of the research remains fragmented, we conduct an in-depth study of the advanced work from a new perspective of learning the part relationship. In this perspective, we first consolidate recent research and benchmark syntheses with new taxonomies. Based on this consolidation, we revisit the universal challenges in fine-grained part segmentation and recognition tasks and propose new solutions by part relationship learning for these important challenges. Furthermore, we conclude several promising lines of research in fine-grained visual parsing for future research.
