3D Point Cloud Generative Adversarial Network Based on Tree Structured Graph Convolutions

Shu, Dongwook; Park, Sung Woo; Kwon, Junseok

doi:10.1109/iccv.2019.00396

Cited by 233 publications

(242 citation statements)

References 35 publications

(35 reference statements)

Supporting

Mentioning

242

Contrasting

Order By: Relevance

“…To overcome the redundancy and structural irregularity of point samples, Ramasinghe et al [2019] propose Spectral-GANs to synthesize shapes using a spherical-harmonicsbased representation. Shu et al [2019] propose tree-GAN to perform graph convolutions in a tree and recently extend it into a multi-rooted version. Hui et al [2020] design a progressive deconvolution network to generate 3D point clouds, while Arshad et al…”

Section: Related Workmentioning

confidence: 99%

“…Table 1. Quantitative comparison of the generated shapes produced by SP-GAN and five state-of-the-art methods, i.e., r-GAN [Achlioptas et al 2018], tree-GAN [Shu et al 2019], PointFlow [Yang et al 2019], PDGN [Hui et al 2020], and ShapeGF [Cai et al 2020]. We follow the same settings to conduct this experiment as in the state-of-the-art methods.…”

Section: Part Co-segmentationmentioning

confidence: 99%

“…Another feasible solution is to design deep generative models to characterize the shape distribution and learn to directly generate shapes in an unsupervised manner. Prior researches generate shapes represented as 3D point clouds, typically by learning to map random latent code to point clouds, via auto-regressive models [Sun et al 2020], flow-based models [Kim et al 2020;Klokov et al 2020;Yang et al 2019], and generative adversarial nets (GAN) [Achlioptas et al 2018;Hui et al 2020;Shu et al 2019]. Though substantial progress has been made, the problem is still very challenging, due to the diverse shape variations and the high complexity in 3D space.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Sp-Gan

Hui

2021

ACM Trans. Graph.

View full text Add to dashboard Cite

We present SP-GAN, a new unsupervised sphere-guided generative model for direct synthesis of 3D shapes in the form of point clouds. Compared with existing models, SP-GAN is able to synthesize diverse and high-quality shapes with fine details and promote controllability for part-aware shape generation and manipulation, yet trainable without any parts annotations. In SP-GAN, we incorporate a global prior (uniform points on a sphere) to spatially guide the generative process and attach a local prior (a random latent code) to each sphere point to provide local details. The key insight in our design is to disentangle the complex 3D shape generation task into a global shape modeling and a local structure adjustment, to ease the learning process and enhance the shape generation quality. Also, our model forms an implicit dense correspondence between the sphere points and points in every generated shape, enabling various forms of structure-aware shape manipulations such as part editing, part-wise shape interpolation, and multi-shape part composition, etc., beyond the existing generative models. Experimental results, which include both visual and quantitative evaluations, demonstrate that our model is able to synthesize diverse point clouds with fine details and less noise, as compared with the state-of-the-art models.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Part Co-segmentationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Sp-Gan

Hui

2021

ACM Trans. Graph.

View full text Add to dashboard Cite

show abstract

“…These examples are generated using deep learning or simple geometrical manipulation to create synthetic datasets. GAN networks proposed to generate the synthetic 3D object point cloud are relevant for object detection and recognition (Shu et al, 2019). Existing frameworks lack the generation of synthetic data for large scale 3D point cloud.…”

Section: Deep Learning Networkmentioning

confidence: 99%

Simulation-Based Data Augmentation Using Physical Priors for Noise Filtering Deep Neural Network

Jameela

Chen

Sit

et al. 2020

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. LiDAR (Light Detection and Ranging) mounted with static and mobile vehicles has been rapidly adopted as a primary sensor for mapping natural and built environments for a range of civil and military applications. Recently, technology advancement in electro-optical engineering enables acquiring laser returns at high pulse repetition frequency (PRF) from 100Hz to 2MHz for airborne LiDAR, which leads to an increase in the density of 3D point cloud significantly. Traditional systems with lower PRF had a single pulse-in-air zone (PIA) big enough to avoid a mismatch between pulse pair at the receiver. Modern multiple pulses-in-air (MPIA) technology ensures multiple windows of operational ranges for single flight line and no blind-zones; downside of the technology is projection of atmospheric returns closer to same PIA zone of neighbouring ground points and more likely to be overlapping with objects of interest. These characteristics of noise compromise the quality of the scene and encourage usage of noise filtering neural network as existing filters are not effective. A noise filtering deep neural network requires a considerable volume of the diverse annotated dataset, which is expensive. We developed simulation for data augmentation based on physical priors and Gaussian generative function. Our study compares deep learning networks for noise filtering and shows performance gain on 3D U-Net. Then, we evaluate 3D U-Net for simulation-based data augmentation, which shows an increase in precision and F1-score. We also provide an analysis of the underline spatial distribution of points and their impact on data augmentation, and noise filtering.

show abstract

“…Point clouds can capture a much higher resolution than voxels, and can be processed using simpler manipulations than meshes. By leveraging the flexibility of deep learning, a deep generative model of point clouds enables a variety of synthesis tasks such as generation, reconstruction, and super-resolution [1,2,13,18,26,36,39,43,47]. Because it is difficult to measure the quality of a generated point cloud numerically, most studies employ flow-based generative models [6,10,20] or generative adversarial networks (GANs) [9].…”

mentioning

confidence: 99%

ChartPointFlow for Topology-Aware 3D Point Cloud Generation

Kimura

Matsubara

Uehara

2021

Proceedings of the 29th ACM International Conference on Multimedia

View full text Add to dashboard Cite

A point cloud serves as a representation of the surface of a threedimensional (3D) shape. Deep generative models have been adapted to model their variations typically using a map from a ball-like set of latent variables. However, previous approaches did not pay much attention to the topological structure of a point cloud, despite that a continuous map cannot express the varying numbers of holes and intersections. Moreover, a point cloud is often composed of multiple subparts, and it is also difficult to express. In this study, we propose ChartPointFlow, a flow-based generative model with multiple latent labels for 3D point clouds. Each label is assigned to points in an unsupervised manner. Then, a map conditioned on a label is assigned to a continuous subset of a point cloud, similar to a chart of a manifold. This enables our proposed model to preserve the topological structure with clear boundaries, whereas previous approaches tend to generate blurry point clouds and fail to generate holes. The experimental results demonstrate that ChartPointFlow achieves state-of-the-art performance in terms of generation and reconstruction compared with other point cloud generators. Moreover, ChartPointFlow divides an object into semantic subparts using charts, and it demonstrates superior performance in case of unsupervised segmentation. CCS CONCEPTS• Computing methodologies → Point-based models.

show abstract

3D Point Cloud Generative Adversarial Network Based on Tree Structured Graph Convolutions

Cited by 233 publications

References 35 publications

Sp-Gan

Sp-Gan

Simulation-Based Data Augmentation Using Physical Priors for Noise Filtering Deep Neural Network

ChartPointFlow for Topology-Aware 3D Point Cloud Generation

Contact Info

Product

Resources

About