PFCNN: Convolutional Neural Networks on 3D Surfaces Using Parallel Frames

Yang, Yuqi; Liu, Shilin; Pan, Hao; Liu, Yang; Tong, Xin

doi:10.1109/cvpr42600.2020.01359

Cited by 34 publications

(19 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Generally, those state-ofthe-art methods submitted to the testing server are classified mainly by their convolution categories and input data types. As shown in Table 5, some of the networks such as 3DMV [63], PFCNN [64], and Tangent Convolution [65] retrieved semantic labels from 2D and 3D information, while the other networks listed below were trained with only 3D input data and can be divided into pointwise convolution and graph convolution. For convincing comparison, we adjusted the depth of the IAG-MLP network to ensure comparable capacity of the IAGC model with the point convolution baseline.…”

Section: Results On the Scannet Datasetmentioning

confidence: 99%

IAGC: Interactive Attention Graph Convolution Network for Semantic Segmentation of Point Clouds in Building Indoor Environment

Zhai

Zou

et al. 2022

IJGI

View full text Add to dashboard Cite

Point-based networks have been widely used in the semantic segmentation of point clouds owing to the powerful 3D convolution neural network (CNN) baseline. Most of the current methods resort to intermediate regular representations for reorganizing the structure of point clouds for 3D CNN networks, but they may neglect the inherent contextual information. In our work, we focus on capturing discriminative features with the interactive attention mechanism and propose a novel method consisting of the regional simplified dual attention network and global graph convolution network. Firstly, we cluster homogeneous points into superpoints and construct a superpoint graph to effectively reduce the computation complexity and greatly maintain spatial topological relations among superpoints. Secondly, we integrate cross-position attention and cross-channel attention into a single head attention module and design a novel interactive attention gating (IAG)-based multilayer perceptron (MLP) network (IAG–MLP), which is utilized for the expansion of the receptive field and augmentation of discriminative features in local embeddings. Afterwards, the combination of stacked IAG–MLP blocks and the global graph convolution network, called IAGC, is proposed to learn high-dimensional local features in superpoints and progressively update these local embeddings with the recurrent neural network (RNN) network. Our proposed framework is evaluated on three indoor open benchmarks, and the 6-fold cross-validation results of the S3DIS dataset show that the local IAG–MLP network brings about 1% and 6.1% improvement in overall accuracy (OA) and mean class intersection-over-union (mIoU), respectively, compared with the PointNet local network. Furthermore, our IAGC network outperforms other CNN-based approaches in the ScanNet V2 dataset by at least 7.9% in mIoU. The experimental results indicate that the proposed method can better capture contextual information and achieve competitive overall performance in the semantic segmentation task.

show abstract

Section: Results On the Scannet Datasetmentioning

confidence: 99%

IAGC: Interactive Attention Graph Convolution Network for Semantic Segmentation of Point Clouds in Building Indoor Environment

Zhai

Zou

et al. 2022

IJGI

View full text Add to dashboard Cite

show abstract

“…Parallel transport is a choice motivated by differential geometry [Pan et al 2018], which can be combined with circular harmonics [Wiersma et al 2020] or pooling over multiple coordinates [Poulenard and Ovsjanikov 2018] to avoid dependence on a local coordinate system. Yang et al [2020] employ locally flat connections for a similar purpose.…”

Section: Neural Network On Meshesmentioning

confidence: 99%

HodgeNet

Smirnov

Solomon

2021

ACM Trans. Graph.

View full text Add to dashboard Cite

Fig. 1. Mesh segmentation results on the full-resolution MIT animation dataset. Each mesh in the dataset contains 20,000 faces (10,000 vertices). We show an example ground truth segmentation in the bottom-left. In contrast to previous works, which downsample each mesh by more than 10×, we efficiently process dense meshes both at train and test time.Constrained by the limitations of learning toolkits engineered for other applications, such as those in image processing, many mesh-based learning algorithms employ data flows that would be atypical from the perspective of conventional geometry processing. As an alternative, we present a technique for learning from meshes built from standard geometry processing modules and operations. We show that low-order eigenvalue/eigenvector computation from operators parameterized using discrete exterior calculus is amenable to efficient approximate backpropagation, yielding spectral per-element or per-mesh features with similar formulas to classical descriptors like the heat/wave kernel signatures. Our model uses few parameters, generalizes to high-resolution meshes, and exhibits performance and time complexity on par with past work.

show abstract

“…Some works project local neighborhoods into tangent planes and process them with 2D convolutions. The tangent plane parameters can be found using point tangent estimation [27], or approximated [9,15,35]. The downside of these approaches is that they lose the information of 1 dimension given that they project the points to a local 2D plane.…”

Section: Related Workmentioning

confidence: 99%

“…Other approaches use multilayer perceptrons (MLP) to process point clouds directly [19,20,29]. A third approach is to project the points to an intermediate grid structure where 2D convolutions can be used [9,15,35]. Lately, with the success of transformers and attention mechanisms in the area of natural language processing (NLP) [30], these methods are starting to be used for 3D point cloud problems [6,21].…”

Section: Introductionmentioning

confidence: 99%

Two Heads are Better than One: Geometric-Latent Attention for Point Cloud Classification and Segmentation

Cuevas-Velasquez¹,

Gallego²,

Fisher³

2021

Preprint

View full text Add to dashboard Cite

We present an innovative two-headed attention layer that combines geometric and latent features to segment a 3D scene into semantically meaningful subsets. Each head combines local and global information, using either the geometric or latent features, of a neighborhood of points and uses this information to learn better local relationships. This Geometric-Latent attention layer (Ge-Latto) is combined with a sub-sampling strategy to capture global features. Our method is invariant to permutation thanks to the use of shared-MLP layers, and it can also be used with point clouds with varying densities because the local attention layer does not depend on the neighbor order. Our proposal is simple yet robust, which allows it to achieve competitive results in the ShapeNetPart and ModelNet40 datasets, and the state-of-the-art when segmenting the complex dataset S3DIS, with 69.2% IoU on Area 5, and 89.7% overall accuracy using K-fold crossvalidation on the 6 areas.

show abstract

PFCNN: Convolutional Neural Networks on 3D Surfaces Using Parallel Frames

Cited by 34 publications

References 30 publications

IAGC: Interactive Attention Graph Convolution Network for Semantic Segmentation of Point Clouds in Building Indoor Environment

IAGC: Interactive Attention Graph Convolution Network for Semantic Segmentation of Point Clouds in Building Indoor Environment

HodgeNet

Two Heads are Better than One: Geometric-Latent Attention for Point Cloud Classification and Segmentation

Contact Info

Product

Resources

About