PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers

  title={PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers},
  author={Xumin Yu and Yongming Rao and Ziyi Wang and Zuyan Liu and Jiwen Lu and Jie Zhou},
  journal={2021 IEEE/CVF International Conference on Computer Vision (ICCV)},
  • Xumin YuYongming Rao Jie Zhou
  • Published 19 August 2021
  • Computer Science
  • 2021 IEEE/CVF International Conference on Computer Vision (ICCV)
Point clouds captured in real-world applications are of-ten incomplete due to the limited sensor resolution, single viewpoint, and occlusion. Therefore, recovering the complete point clouds from partial ones becomes an indispensable task in many practical applications. In this paper, we present a new method that reformulates point cloud completion as a set-to-set translation problem and design a new model, called PoinTr that adopts a transformer encoder-decoder architecture for point cloud… 

AdaPoinTr: Diverse Point Cloud Completion with Adaptive Geometry-Aware Transformers

A new method that reformulates point cloud completion as a set-to-set translation problem and design a new model, called PoinTr, which adopts a Transformer encoder-decoder architecture for point cloudpletion, which attains promising performance and can be extended to the scene-level point clouds completion scenario by designing a new geometry-enhanced semantic scene completion framework.

CompleteDT: Point Cloud Completion with Dense Augment Inference Transformers

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.

Partial-to-Partial Point Generation Network for Point Cloud Completion

A novel Partial-to-Partial Point Generation Network (GNet) is proposed, a learning-based approach for point cloud completion that outperforms state-of-the-art point clouds completion methods.

PDP-NET: Patch-Based Dual-Path Network for Point Cloud Completion

A novel Patch-based Dual-Path Network (PDP-Net) for point cloud completion is proposed, which leverages the advantages of different encoder architectures, with one path providing estimation for the global structure of the missing part, and the other path filling in the details by generating several point cloud patches.

Multi-Dimensional Graph Interactional Network for Progressive Point Cloud Completion

This work develops a recursive point cloud pyramid decoder (RPPD) for generating multistage completed point clouds progressively, which incorporates extra feedback connections into the bottom-up backbone layers and designs a depth map discriminator combined with differentiable rendering to match the distribution of generated and real point clouds.

Point Cloud Completion via Multi-Scale Edge Convolution and Attention

This work proposes a novel point cloud completion network based on multi-scale edge convolution and attention mechanism, named MEAPCN, which achieves a superior completion performance with much smaller model size and much lower computation cost.

Self-Supervised Point Cloud Completion on Real Traffic Scenes Via Scene-Concerned Bottom-Up Mechanism

This paper proposes a self-supervised point cloud completion method (TraPCC) for vehicles in real traffic scenes without any complete data, based on the symmetry and similarity of vehicles, and designs a bottom-up mechanism to focus on both local geometry details and global shape features of inputs.

AGFA-Net: Adaptive Global Feature Augmentation Network for Point Cloud Completion

Experimental results and analyses demonstrate that the proposed adaptive global feature augmentation network (AGFA-Net) can generate finer shapes of point clouds and outperforms other state-of-the-art methods under widely used benchmark point completion network (PCN) dataset and several terrestrial laser scanning data.

Cyclic Global Guiding Network for Point Cloud Completion

A cyclic global guiding network structure is proposed and applied to point cloud completion tasks and uses the relationship between the point and the plane to increase the compatibility between the network learning and the original sparse point cloud.

Learning a Structured Latent Space for Unsupervised Point Cloud Completion

This work proposes a novel framework, which learns a unified and structured latent space that encoding both partial and complete point clouds, and adopts a series of constraints including structured ranking regularization, latent code swapping constraint, and distribution supervision on the related partial point clouds.



PMP-Net: Point Cloud Completion by Learning Multi-step Point Moving Paths

A novel neural network, named PMP-Net, is designed to mimic the behavior of an earth mover, which predicts a unique point moving path for each point according to the constraint of total point moving distances.

GRNet: Gridding Residual Network for Dense Point Cloud Completion

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

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.

Cascaded Refinement Network for Point Cloud Completion

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

Qualitative and quantitative evaluations demonstrate that the proposed network outperforms current state-of-the art methods especially on detail preservation.

Dense 3D Point Cloud Reconstruction Using a Deep Pyramid Network

This work introduces DensePCR, a deep pyramidal network for point cloud reconstruction that hierarchically predicts point clouds of increasing resolution, and proposes an architecture that first predicts a low-resolution point cloud, and then hierarchically increases the resolution by aggregating local and global point features to deform a grid.

PF-Net: Point Fractal Network for 3D Point Cloud Completion

A Point Fractal Network (PF-Net), a novel learning-based approach for precise and high-fidelity point cloud completion that preserves the spatial arrangements of the incomplete point cloud and can figure out the detailed geometrical structure of the missing region(s) in the prediction.

Morphing and Sampling Network for Dense Point Cloud Completion

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).

Dynamic Graph CNN for Learning on Point Clouds

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.

FoldingNet: Interpretable Unsupervised Learning on 3D Point Clouds

A novel end-to-end deep auto-encoder is proposed to address unsupervised learning challenges on point clouds, and this decoder structure is shown, in theory, to be a generic architecture that is able to reconstruct an arbitrary point cloud from a 2D grid.