Variational Relational Point Completion Network

@article{Pan2021VariationalRP,
  title={Variational Relational Point Completion Network},
  author={Liang Pan and Xinyi Chen and Zhongang Cai and Junzhe Zhang and Haiyu Zhao and Shuai Yi and Ziwei Liu},
  journal={2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2021},
  pages={8520-8529}
}
  • Liang Pan, Xinyi Chen, Ziwei Liu
  • Published 20 April 2021
  • Computer Science
  • 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Real-scanned point clouds are often incomplete due to viewpoint, occlusion, and noise. Existing point cloud completion methods tend to generate global shape skeletons and hence lack fine local details. Furthermore, they mostly learn a deterministic partial-to-complete mapping, but overlook structural relations in man-made objects. To tackle these challenges, this paper proposes a variational framework, Variational Relational point Completion network (VRC-Net) with two appealing properties: 1… 
A Conditional Point Diffusion-Refinement Paradigm for 3D Point Cloud Completion
TLDR
This work proposes a novel Point Diffusion-Refinement (PDR) paradigm for point cloud completion that can effectively and efficiently extract multi-level features from partially observed point clouds to guide completion, and accurately manipulate spatial locations of 3D points to obtain smooth surfaces and sharp details.
Point cloud completion via structured feature maps using a feedback network
TLDR
This paper proposes FSNet, a feature structuring module that can adaptively aggregate point-wise features into a 2D structured feature map by learning multiple latent patterns from local regions, and integrates FSNet into a coarse-tofine pipeline for point cloud completion.
Reconstruction-Aware Prior Distillation for Semi-supervised Point Cloud Completion
TLDR
Extensive experiments demonstrate that RaPD, the first semi-supervised point cloud completion method, achieves superior performance to previous methods on both homologous and heterologous scenarios.
CompleteDT: Point Cloud Completion with Dense Augment Inference Transformers
TLDR
A novel point cloud completion network, CompleteDT, which is based on the transformer, which largely outperforms the state-of-the-art methods and can learn features within neighborhoods and explore the relationship among these neighborhoods.
Temporal Point Cloud Completion With Pose Disturbance
TLDR
This work proposes a point cloud completion framework that accepts a sequence of unaligned and sparse inputs, and outputs consistent and aligned point clouds and is the first to utilize temporal information and ensure temporal consistency with limited transformation.
PMP-Net++: Point Cloud Completion by Transformer-Enhanced Multi-step Point Moving Paths
TLDR
A novel neural network is designed, named PMP-Net++, to mimic behavior of an earth mover, which improves quality of predicted complete shape in point cloud completion and introduces a transformer-enhanced representation learning network, which significantly improves completion performance.
Point cloud completion on structured feature map with feedback network
TLDR
This paper proposes FSNet, a feature structuring module that can adaptively aggregate point-wise features into a 2D structured feature map by learning multiple latent patterns from local regions and integrates FSNet into a coarse-to-fine pipeline for point cloud completion.
PointAttN: You Only Need Attention for Point Cloud Completion
TLDR
The proposed framework, namely PointAttN, is simple, neat and effective, which can precisely capture the structural information of 3D shapes and predict complete point clouds with highly detailed geometries and outperforms state-of-the-art methods by a large margin.
Learning Local Displacements for Point Cloud Completion
TLDR
A novel approach aimed at object and semantic scene completion from a partial scan represented as a 3D point cloud is proposed and a second model is introduced that assembles the authors' layers within a transformer architecture, achieving state-of-the-art performance.
ASFM-Net: Asymmetrical Siamese Feature Matching Network for Point Completion
TLDR
The Siamese auto-encoder neural network is adopted to map the partial and complete input point cloud into a shared latent space, which can capture detailed shape prior and an iterative refinement unit is designed to generate complete shapes with fine-grained details by integrating prior information.
...
...

References

SHOWING 1-10 OF 32 REFERENCES
Cascaded Refinement Network for Point Cloud Completion
TLDR
This work proposes a cascaded refinement network together with a coarse-to-fine strategy to synthesize the detailed object shapes and designs a patch discriminator that guarantees every local area has the same pattern with the ground truth to learn the complicated point distribution.
Detail Preserved Point Cloud Completion via Separated Feature Aggregation
TLDR
Qualitative and quantitative evaluations demonstrate that the proposed network outperforms current state-of-the art methods especially on detail preservation.
PCN: Point Completion Network
TLDR
The experiments show that PCN produces dense, complete point clouds with realistic structures in the missing regions on inputs with various levels of incompleteness and noise, including cars from LiDAR scans in the KITTI dataset.
Morphing and Sampling Network for Dense Point Cloud Completion
TLDR
This work proposes a novel approach to complete the partial point cloud in two stages, which outperforms the existing methods in both the Earth Mover's Distance (EMD) and the Chamfer Distance (CD).
Point Cloud Completion by Skip-Attention Network With Hierarchical Folding
TLDR
This work proposes Skip-Attention Network (SA-Net), a skip-attention mechanism to effectively exploit the local structure details of incomplete point clouds during the inference of missing parts, and proposes a novel structure-preserving decoder with hierarchical folding for complete shape generation.
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.
GRNet: Gridding Residual Network for Dense Point Cloud Completion
TLDR
This work devise two novel differentiable layers, named Gridding and Gridding Reverse, to convert between point clouds and 3D grids without losing structural information, and presents the differentiable Cubic Feature Sampling layer to extract features of neighboring points, which preserves context information.
TopNet: Structural Point Cloud Decoder
TLDR
This work proposes a novel decoder that generates a structured point cloud without assuming any specific structure or topology on the underlying point set, and significantly outperforms state-of-the-art 3D point cloud completion methods on the Shapenet dataset.
FoldingNet: Point Cloud Auto-Encoder via Deep Grid Deformation
TLDR
A novel end-to-end deep auto-encoder is proposed to address unsupervised learning challenges on point clouds, and is shown, in theory, to be a generic architecture that is able to reconstruct an arbitrary point cloud from a 2D grid.
ECG: Edge-aware Point Cloud Completion with Graph Convolution
  • Liang Pan
  • Computer Science
    IEEE Robotics and Automation Letters
  • 2020
TLDR
The proposed ECG - an Edge-aware point cloud Completion network with Graph convolution, which facilitates fine-grained 3D point cloud shape generation with multi-scale edge features, significantly outperforms previous state-of-the-art (SOTA) methods for point cloud completion.
...
...