Monte Carlo convolution for learning on non-uniformly sampled point clouds

Hermosilla, Pedro; Ritschel, Tobias; Vázquez, Pere‐Pau; Vinacua, Àlvar; Ropinski, Timo

doi:10.1145/3272127.3275110

Cited by 233 publications

(201 citation statements)

References 34 publications

Supporting

Mentioning

190

Contrasting

Unclassified

Order By: Relevance

“…Beyond learning on unstructured point clouds, there have been some notable extension works, such as learning with hierarchical structures [28,14,35,36], learning with selforganizing network [19], learning to map a 3D point cloud to a 2D grid [43,8], addressing large-scale point cloud segmentation [15], handling non-uniform point cloud [11], and employing spectral analysis [45]. Such ideas are orthogonal to our method, and adding them on top of our proposed convolution could be an interesting future research.…”

Section: Related Workmentioning

confidence: 99%

ShellNet: Efficient Point Cloud Convolutional Neural Networks Using Concentric Shells Statistics

Zhang

Hua

Yeung

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

319

187

View full text Add to dashboard Cite

Deep learning with 3D data has progressed significantly since the introduction of convolutional neural networks that can handle point order ambiguity in point cloud data. While being able to achieve good accuracies in various scene understanding tasks, previous methods often have low training speed and complex network architecture. In this paper, we address these problems by proposing an efficient endto-end permutation invariant convolution for point cloud deep learning. Our simple yet effective convolution operator named ShellConv uses statistics from concentric spherical shells to define representative features and resolve the point order ambiguity, allowing traditional convolution to perform on such features. Based on ShellConv we further build an efficient neural network named ShellNet to directly consume the point clouds with larger receptive fields while maintaining less layers. We demonstrate the efficacy of ShellNet by producing state-of-the-art results on object classification, object part segmentation, and semantic scene segmentation while keeping the network very fast to train. Our code is publicly available in our project page 1 .

show abstract

Section: Related Workmentioning

confidence: 99%

ShellNet: Efficient Point Cloud Convolutional Neural Networks Using Concentric Shells Statistics

Zhang

Hua

Yeung

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

319

187

View full text Add to dashboard Cite

show abstract

“…More recently, some attempts have been made to design a convolution that operates directly on points [2,45,20,14,13]. These methods use the spatial localization property of Figure 1.…”

Section: Introductionmentioning

confidence: 99%

KPConv: Flexible and Deformable Convolution for Point Clouds

Thomas¹,

Qi²,

Deschaud³

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

2,128

1,606

View full text Add to dashboard Cite

We present Kernel Point Convolution 1 (KPConv), a new design of point convolution, i.e. that operates on point clouds without any intermediate representation. The convolution weights of KPConv are located in Euclidean space by kernel points, and applied to the input points close to them. Its capacity to use any number of kernel points gives KP-Conv more flexibility than fixed grid convolutions. Furthermore, these locations are continuous in space and can be learned by the network. Therefore, KPConv can be extended to deformable convolutions that learn to adapt kernel points to local geometry. Thanks to a regular subsampling strategy, KPConv is also efficient and robust to varying densities. Whether they use deformable KPConv for complex tasks, or rigid KPconv for simpler tasks, our networks outperform state-of-the-art classification and segmentation approaches on several datasets. We also offer ablation studies and visualizations to provide understanding of what has been learned by KPConv and to validate the descriptive power of deformable KPConv. AbstractThis supplementary document is organized as follows:• Sec. A details our network architectures, the training parameters, and compares the model sizes and speeds.• Sec. B presents the kernel point initialization method.

show abstract

“…As for now, employing different parameter settings for different datasets is practically more feasible. Many recent deep neural network models for point cloud processing, such as MCCNN [14], also trained separate models for different datasets. A potential solution is domain adaptation [2], especially multi-source domain adaptation, that has proven beneficial for learning source data from different domains [12].…”

Section: Generalizabilitymentioning

confidence: 99%

“…• A first and foremost work is to update our backbone network to stateof-the-art deep learning models for processing point clouds. For example, Hermosilla et al proposed MCCNN that utilizes Monte Carlo up-and down-sampling to preserve the original sample density, making it more suitable for non-uniformly distributed point clouds [14]. We consider MCCNN as an important future improvement to enhance the effectiveness and robustness of LassoNet.…”

Section: Limitations and Future Workmentioning

confidence: 99%

LassoNet: Deep Lasso-Selection of 3D Point Clouds

Chen

Zeng

Yang

et al. 2020

IEEE Trans. Visual. Comput. Graphics

View full text Add to dashboard Cite

Fig. 1. LassoNet enables effective lasso-selection of 3D point clouds based on a latent mapping from viewpoint and lasso to target point clouds. LassoNet is particularly efficient for selecting multiple regions (insets 1, 2, 3) in a complex scene (left), since no viewpoint changing is required to select occluded points. Notice here the insets are viewed from different viewpoints.Abstract-Selection is a fundamental task in exploratory analysis and visualization of 3D point clouds. Prior researches on selection methods were developed mainly based on heuristics such as local point density, thus limiting their applicability in general data. Specific challenges root in the great variabilities implied by point clouds (e.g., dense vs. sparse), viewpoint (e.g., occluded vs. non-occluded), and lasso (e.g., small vs. large). In this work, we introduce LassoNet, a new deep neural network for lasso selection of 3D point clouds, attempting to learn a latent mapping from viewpoint and lasso to point cloud regions. To achieve this, we couple user-target points with viewpoint and lasso information through 3D coordinate transform and naive selection, and improve the method scalability via an intention filtering and farthest point sampling. A hierarchical network is trained using a dataset with over 30K lasso-selection records on two different point cloud data. We conduct a formal user study to compare LassoNet with two state-of-the-art lasso-selection methods. The evaluations confirm that our approach improves the selection effectiveness and efficiency across different combinations of 3D point clouds, viewpoints, and lasso selections. Project Website: https://lassonet.github.io

show abstract

Monte Carlo convolution for learning on non-uniformly sampled point clouds

Cited by 233 publications

References 34 publications

ShellNet: Efficient Point Cloud Convolutional Neural Networks Using Concentric Shells Statistics

ShellNet: Efficient Point Cloud Convolutional Neural Networks Using Concentric Shells Statistics

KPConv: Flexible and Deformable Convolution for Point Clouds

LassoNet: Deep Lasso-Selection of 3D Point Clouds

Contact Info

Product

Resources

About