Multiresolution Tree Networks for 3D Point Cloud Processing

By Matheus Gadelha, Rui Wang, Subhransu Maji. 2018

271 citations recorded by the source corpus.

We present multiresolution tree-structured networks to process point clouds for 3D shape understanding and generation tasks. Our network represents a 3D shape as a set of locality-preserving 1D ordered list of points at multiple resolutions. This allows efficient feed-forward processing through 1D convolutions, coarse-to-fine analysis through a multi-grid architecture, and it leads to faster convergence and small memory footprint during training. The proposed tree-structured encoders can be used to classify shapes and outperform existing point-based architectures on shape classification benchmarks, while tree-structured decoders can be used for generating point clouds directly and they outperform existing approaches for image-to-shape inference tasks learned using the ShapeNet dataset. Our model also allows unsupervised learning of point-cloud based shapes by using a variational autoencoder, leading to higher-quality generated shapes.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Geology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Vision and Pattern Recognition

In Geology

Citations in the corpus

Learning Implicit Fields for Generative Shape Modeling (2018, background): “There have been a variety of 3D shape representations for deep learning of shapes, such as voxel grids [9, 14, 28, 45, 46], octrees [18, 35, 40, 43, 44], multi-view images [29, 39], point clouds [1, 12, 13, 32, 33, 47, 48], geometry images [37, 38], deformable mesh/patches [16, 38, 42, 47], and part”
Diffusion Probabilistic Models for 3D Point Cloud Generation (2021, background): “Point Cloud Generation Early point cloud generation methods [1, 7] treat point clouds as N × 3 matrices, where N is the fixed number of points, converting the point cloud generation problem to a matrix generation problem, so that existing generative models are readily applicable.”
Learning Gradient Fields for Shape Generation (2020, neutral): “N×3 where N is predefined) [2,18,67,56,68,17,52,59], or relies on heuristic set distance functions such as Chamfer distance and Earth Mover Distance [25,65,19,12,6].”
Self-Supervised Pretraining of 3D Features on any Point-Cloud (2021, methodology): “Most methods on self-supervised learning focus on single 3D object representation with different applications to reconstruction, classification or part segmentation [1, 2, 25, 33, 35, 44, 49, 74, 99, 110].”
PF-Net: Point Fractal Network for 3D Point Cloud Completion (2020, methodology): “To evaluate the methods mentioned above, we use the evaluation metric by [37, 38].”
PointContrast: Unsupervised Pre-training for 3D Point Cloud Understanding (2020, methodology): “Second, is the choice of neighborhood aggregation mechanism which could either be hierarchical [48, 33, 82, 16, 35], spatial CNN-like [30, 73, 37, 85, 59], spectral [78, 62, 65] or graph-based [72, 64, 68, 54].”
BSP-Net: Generating Compact Meshes via Binary Space Partitioning (2019, methodology): “Amongst these we find a category of techniques pioneered by PointNet [33] that express surfaces as point clouds [1, 9, 10, 33, 34, 53, 56], and techniques pioneered by AtlasNet [16] that adopt a 2D-to-3D mapping process [47, 39, 44, 53].”
Relation-Shape Convolutional Neural Network for Point Cloud Analysis (2019, methodology): “Some works transform point cloud to regular voxels [42, 22, 3] or multi-view images [35, 2, 5] for easy application of classic grid CNN.”
Patch-Based Progressive 3D Point Set Upsampling (2018, background): “A few recent works have exploited the power of multiscale representation [12, 24, 28, 37, 49] and skipconnection [8,43] in 3D learning.”
TopNet: Structural Point Cloud Decoder (2019, background): “[10] represents a point cloud as a 1D list of spatially ordered points, and generates a point cloud through a tree network in which each branch aims at representing different resolutions of the point cloud which are connected through multiresolution convolution.”
DensePoint: Learning Densely Contextual Representation for Efficient Point Cloud Processing (2019, methodology): “A straightforward strategy is to transform point cloud into regular voxels [52, 27, 4] or multi-view images [43, 3, 6], for easy application of CNN.”
Global-Local Bidirectional Reasoning for Unsupervised Representation Learning of 3D Point Clouds (2020, methodology): “Self-Reconstruction: Self-reconstruction, or point autoencoding, is a widely used technique for unsupervised point cloud representation learning [1,8,14,26,47,55,56].”
Image-Based 3D Object Reconstruction: State-of-the-Art and Trends in the Deep Learning Era (2019, methodology): “Compared to standard encoders, 3D-VAE can be used to randomly sample from the latent space, to generate variations of an input, and to reconstruct multiple plausible 3D shapes from an input image [21], [22].”
Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts (2020, methodology): “Following suit, recent works attempt to adapt the 2D pretext tasks to 3D, but mostly focus on single object classification tasks on ShapeNet [1, 17, 69, 22, 37, 62, 26, 51, 50].”