A Closer Look at Local Aggregation Operators in Point Cloud Analysis

@inproceedings{Liu2020ACL,
  title={A Closer Look at Local Aggregation Operators in Point Cloud Analysis},
  author={Ze Liu and Han Hu and Yue Cao and Zheng Zhang and Xin Tong},
  booktitle={ECCV},
  year={2020}
}
Recent advances of network architecture for point cloud processing are mainly driven by new designs of local aggregation operators. However, the impact of these operators to network performance is not carefully investigated due to different overall network architecture and implementation details in each solution. Meanwhile, most of operators are only applied in shallow architectures. In this paper, we revisit the representative local aggregation operators and study their performance using the… 
Differentiable Convolution Search for Point Cloud Processing
TLDR
PointSeaConv is proposed, a novel differential convolution search paradigm on point clouds that can work in a purely data-driven manner and thus is capable of auto-creating a group of suitable convolutions for geometric shape modeling and introduces epsilon-greedy algorithm to alleviate the effect of discretization error.
Towards Efficient Point Cloud Graph Neural Networks Through Architectural Simplification
TLDR
It is found that it is possible to radically simplify graph neural network models so long as the feature extraction layer is retained with minimal degradation to model performance; further, it is discovered that it was possible to improve performance overall on ModelNet40 and S3DIS by improving the design of the feature extractor.
DRINet: A Dual-Representation Iterative Learning Network for Point Cloud Segmentation
TLDR
The DRINet work, which serves as the basic network structure for dual-representation learning with great flexibility at feature transferring and less computation cost, especially for large-scale point clouds, is proposed.
CGA-Net: Category Guided Aggregation for Point Cloud Semantic Segmentation
TLDR
This paper presents a customized module, termed as Category Guided Aggregation (CGA), where it first identifies whether the neighbors belong to the same category with the center point or not, and then handles the two types of neighbors with two carefully-designed modules.
Learning Inner-Group Relations on Point Clouds
TLDR
This paper proposes a scalable and efficient module, called group relation aggregator, which computes a feature of a group based on the aggregation of the features of the inner-group points weighted by geometric relations and semantic relations, and adopts this module to design the RPNet.
Exploiting Local Geometry for Feature and Graph Construction for Better 3D Point Cloud Processing with Graph Neural Networks
TLDR
It is demonstrated with multiple challenging benchmarks, with relatively clean CAD models, as well as with real world noisy scans, that the proposed method achieves state of the art results on benchmarks for 3D classification, part segmentation and semantic segmentation.
Dual-Neighborhood Deep Fusion Network for Point Cloud Classification
TLDR
A feature representation learning method, named Dual-Neighborhood Deep Fusion Network (DNDFN), is proposed to serve as an improved point cloud encoder for the task of nonidealized point cloud classification and achieves the state of the arts.
Dual-Neighborhood Deep Fusion Network for Point Cloud Analysis
TLDR
DNDFN, namely, Dual-Neighborhood Deep Fusion Network, is proposed, which can capture the global neighborhood and information transfer convolution (IT-Conv) which can learn the structural information between two points and transfer features through it.
PRA-Net: Point Relation-Aware Network for 3D Point Cloud Analysis
TLDR
A novel framework named Point Relation-Aware Network (PRA-Net), which is composed of an Intra-region Structure Learning (ISL) module and an Inter-region Relation Learning (IRL) modules that can dynamically integrate the local structural information into the point features, and captures inter-region relations adaptively and efficiently.
Adaptive Pyramid Context Fusion for Point Cloud Perception
TLDR
This letter proposes a module, named adaptive pyramid context fusion (APCF), to adaptively capture scales of contextual information from a multiscale feature pyramid for the point cloud, and demonstrates that APCF-Net surpasses its vanilla counterpart by a large margin both in effectiveness and efficiency.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 47 REFERENCES
DensePoint: Learning Densely Contextual Representation for Efficient Point Cloud Processing
TLDR
DensePoint is proposed, a general architecture to learn densely contextual representation for point cloud processing that extends regular grid CNN to irregular point configuration by generalizing a convolution operator, which holds the permutation invariance of points, and achieves efficient inductive learning of local patterns.
Dynamic Graph CNN for Learning on Point Clouds
TLDR
This work proposes a new neural network module suitable for CNN-based high-level tasks on point clouds, including classification and segmentation called EdgeConv, which acts on graphs dynamically computed in each layer of the network.
Relation-Shape Convolutional Neural Network for Point Cloud Analysis
TLDR
RS-CNN, namely, Relation-Shape Convolutional Neural Network, which extends regular grid CNN to irregular configuration for point cloud analysis, achieves the state of the arts and a hierarchical architecture can be developed to achieve contextual shape-aware learning for Point cloud analysis.
PointConv: Deep Convolutional Networks on 3D Point Clouds
TLDR
The dynamic filter is extended to a new convolution operation, named PointConv, which can be applied on point clouds to build deep convolutional networks and is able to achieve state-of-the-art on challenging semantic segmentation benchmarks on 3D point clouds.
A-CNN: Annularly Convolutional Neural Networks on Point Clouds
TLDR
A new method to define and compute convolution directly on 3D point clouds by the proposed annular convolution that can better capture the local neighborhood geometry of each point by specifying the (regular and dilated) ring-shaped structures and directions in the computation.
Interpolated Convolutional Networks for 3D Point Cloud Understanding
TLDR
A novel Interpolated Convolution operation, InterpConv, is proposed to tackle the point cloud feature learning and understanding problem, to utilize a set of discrete kernel weights and interpolate point features to neighboring kernel-weight coordinates by an interpolation function for convolution.
Modeling Local Geometric Structure of 3D Point Clouds Using Geo-CNN
TLDR
The proposed Geo-CNN applies a generic convolution-like operation dubbed as GeoConv to each point and its local neighborhood, which encourages the network to preserve the geometric structure in Euclidean space throughout the feature extraction hierarchy.
PointCNN: Convolution On X-Transformed Points
TLDR
This work proposes to learn an Χ-transformation from the input points to simultaneously promote two causes: the first is the weighting of the input features associated with the points, and the second is the permutation of the points into a latent and potentially canonical order.
Graph Attention Convolution for Point Cloud Semantic Segmentation
TLDR
A novel graph attention convolution, whose kernels can be dynamically carved into specific shapes to adapt to the structure of an object, which can capture the structured features of point clouds for fine-grained segmentation and avoid feature contamination between objects.
Flex-Convolution - Million-Scale Point-Cloud Learning Beyond Grid-Worlds
TLDR
This work introduces a natural generalization flex-convolution of the conventional convolution layer along with an efficient GPU implementation and demonstrates competitive performance on rather small benchmark sets using fewer parameters and lower memory consumption and obtain significant improvements on a million-scale real-world dataset.
...
1
2
3
4
5
...