Filters








6,732 Hits in 7.2 sec

The Devil is in the Pose: Ambiguity-free 3D Rotation-invariant Learning via Pose-aware Convolution [article]

Ronghan Chen, Yang Cong
2022 arXiv   pre-print
Rotation-invariant (RI) 3D deep learning methods suffer performance degradation as they typically design RI representations as input that lose critical global information comparing to 3D coordinates.  ...  local pose in each layer, and the global information can be hierarchically aggregated in the deep networks without extra efforts.  ...  While the features learned by DGCNN [35] change vastly as the shape rotates, our PaRI-Conv can learn consistent representation for corresponding points between point clouds under different rotations.  ... 
arXiv:2205.15210v1 fatcat:ecuvfudqebcl7cqoi4rencbwla

Robust Kernel-based Feature Representation for 3D Point Cloud Analysis via Circular Graph Convolutional Network [article]

Seung Hwan Jung, Minyoung Chung, Yeong-Gil Shin
2021 arXiv   pre-print
Learning discriminative representations of local geometric features is unquestionably the most important task for accurate point cloud analyses.  ...  Moreover, to improve representations of the local descriptors, we propose a global aggregation method. First, we place kernels aligned around each point in the normal direction.  ...  CONCLUSION Encoding rotation-, density-, and scale-robust features is a challenging task for point cloud representation.  ... 
arXiv:2012.12215v4 fatcat:zubhcocyizactixueogmb5hjlu

GraphTER: Unsupervised Learning of Graph Transformation Equivariant Representations via Auto-Encoding Node-wise Transformations [article]

Xiang Gao, Wei Hu, Guo-Jun Qi
2020 arXiv   pre-print
To this end, we propose a novel unsupervised learning of Graph Transformation Equivariant Representations (GraphTER), aiming to capture intrinsic patterns of graph structure under both global and local  ...  In experiments, we apply the learned GraphTER to graphs of 3D point cloud data, and results on point cloud segmentation/classification show that GraphTER significantly outperforms state-of-the-art unsupervised  ...  Also, as features of each node are learned via propagation from transformed and non-transformed nodes isotropically or anisotropically by both local or global sampling, the learned representation is able  ... 
arXiv:1911.08142v2 fatcat:peqs5x3t2nbchec5a7dp6j7jia

GraphTER: Unsupervised Learning of Graph Transformation Equivariant Representations via Auto-Encoding Node-Wise Transformations

Xiang Gao, Wei Hu, Guo-Jun Qi
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
To this end, we propose a novel unsupervised learning of Graph Transformation Equivariant Representations (GraphTER), aiming to capture intrinsic patterns of graph structure under both global and local  ...  In experiments, we apply the learned GraphTER to graphs of 3D point cloud data, and results on point cloud segmentation/classification show that Graph-TER significantly outperforms state-of-the-art unsupervised  ...  Also, as features of each node are learned via propagation from transformed and non-transformed nodes isotropically or anisotropically by both local or global sampling, the learned representation is able  ... 
doi:10.1109/cvpr42600.2020.00719 dblp:conf/cvpr/GaoHQ20 fatcat:227op5s6wnck7ff5wupmnykysu

Learning Inner-Group Relations on Point Clouds [article]

Haoxi Ran, Wei Zhuo, Jun Liu, Li Lu
2021 arXiv   pre-print
In this paper, we explore the possibilities of local relation operators and survey their feasibility. We propose a scalable and efficient module, called group relation aggregator.  ...  We also compare our local aggregator with PointNet++, with around 30% parameters and 50% computation saving.  ...  Our key contributions are manifold: • A novel scalable local aggregator for point clouds is proposed.  ... 
arXiv:2108.12468v1 fatcat:6bg5wux4rnbsbkcty55camkt3u

A Registration-aided Domain Adaptation Network for 3D Point Cloud Based Place Recognition [article]

Zhijian Qiao, Hanjiang Hu, Weiang Shi, Siyuan Chen, Zhe Liu, Hesheng Wang
2021 arXiv   pre-print
To this end, a novel registration-aided 3D domain adaptation network for point cloud based place recognition is proposed.  ...  However, it is time-consuming and effort-costly to obtain high-quality point cloud data for place recognition model training and ground truth for registration in the real world.  ...  However, for the consideration that global feature can be viewed as the aggregation of local features, we incorporate the point cloud registration to help to learn the local point-wise features for three  ... 
arXiv:2012.05018v2 fatcat:qojrmgenfbavlfd6ogwfaawkxe

VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection [article]

Yin Zhou, Oncel Tuzel
2017 arXiv   pre-print
To interface a highly sparse LiDAR point cloud with a region proposal network (RPN), most existing efforts have focused on hand-crafted feature representations, for example, a bird's eye view projection  ...  In this way, the point cloud is encoded as a descriptive volumetric representation, which is then connected to a RPN to generate detections.  ...  Acknowledgement: We are grateful to our colleagues Russ Webb, Barry Theobald, and Jerremy Holland for their valuable input.  ... 
arXiv:1711.06396v1 fatcat:agpuvj5ghvhtjpoys77qllijfi

VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection

Yin Zhou, Oncel Tuzel
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
To interface a highly sparse LiDAR point cloud with a region proposal network (RPN), most existing efforts have focused on hand-crafted feature representations, for example, a bird's eye view projection  ...  In this way, the point cloud is encoded as a descriptive volumetric representation, which is then connected to a RPN to generate detections.  ...  Acknowledgement: We are grateful to our colleagues Russ Webb, Barry Theobald, and Jerremy Holland for their valuable input.  ... 
doi:10.1109/cvpr.2018.00472 dblp:conf/cvpr/ZhouT18 fatcat:4sjzypqxandijkzr4ivivky6ci

Point2SpatialCapsule: Aggregating Features and Spatial Relationships of Local Regions on Point Clouds using Spatial-aware Capsules [article]

Xin Wen, Zhizhong Han, Xinhai Liu, Yu-Shen Liu
2019 arXiv   pre-print
To address this issue, we propose a novel deep learning network, named Point2SpatialCapsule, for aggregating features and spatial relationships of local regions on point clouds, which aims to learn more  ...  Learning discriminative shape representation directly on point clouds is still challenging in 3D shape analysis and understanding.  ...  Recent studies for learning point cloud representations usually involve the following three steps. Each input point cloud is first split into some local regions.  ... 
arXiv:1908.11026v1 fatcat:gekv432wgrcybfskhhgmz766pm

Modeling Local Geometric Structure of 3D Point Clouds using Geo-CNN [article]

Shiyi Lan, Ruichi Yu, Gang Yu, Larry S. Davis
2018 arXiv   pre-print
GeoConv is a generic and efficient operation that can be easily integrated into 3D point cloud analysis pipelines for multiple applications.  ...  Modeling local structure has been proven to be important for the success of convolutional architectures, and researchers exploited the modeling of local point sets in the feature extraction hierarchy.  ...  Acknowledgement The research was partially supported by the Office of Naval Research under Grant N000141612713: Visual Common Sense Reasoning for Multi-agent Activity Prediction and Recognition.  ... 
arXiv:1811.07782v1 fatcat:n7ara5b5lbeirgmnaahd3xlqk4

Semantic Graph Based Place Recognition for 3D Point Clouds [article]

Xin Kong, Xuemeng Yang, Guangyao Zhai, Xiangrui Zhao, Xianfang Zeng, Mengmeng Wang, Yong Liu, Wanlong Li, Feng Wen
2020 arXiv   pre-print
First, we propose a novel semantic graph representation for the point cloud scenes by reserving the semantic and topological information of the raw point cloud.  ...  Unlike most of the existing methods that focus on extracting local, global, and statistical features of raw point clouds, our method aims at the semantic level that can be superior in terms of robustness  ...  Semantic Graph Representation Semantic Segmentation for Point Cloud: Some semantic segmentation methods for point cloud have been proposed recently [24] - [27] .  ... 
arXiv:2008.11459v1 fatcat:fcnzohqpmzfizjcgmi2n3wdtk4

Deep Learning for 3D Point Clouds: A Survey [article]

Yulan Guo, Hanyun Wang, Qingyong Hu, Hao Liu, Li Liu, Mohammed Bennamoun
2020 arXiv   pre-print
To stimulate future research, this paper presents a comprehensive review of recent progress in deep learning methods for point clouds.  ...  However, deep learning on point clouds is still in its infancy due to the unique challenges faced by the processing of point clouds with deep neural networks.  ...  SRINet [60] first projects a point cloud to obtain rotation invariant representations, and then utilizes PointNet-based backbone to extract a global feature and graph-based aggregation to extract local  ... 
arXiv:1912.12033v2 fatcat:qiiyvvuulfccxaiihf2mu23k34

Exploiting Local and Global Structure for Point Cloud Semantic Segmentation with Contextual Point Representations [article]

Xu Wang, Jingming He, Lin Ma
2019 arXiv   pre-print
In this paper, we propose one novel model for point cloud semantic segmentation, which exploits both the local and global structures within the point cloud based on the contextual point representations  ...  local point cloud structure.  ...  National Natural Science Foundation of China (Grant 61871270 and Grant 61672443), in part by the Natural Science Foundation of SZU (grant no. 827000144) and in part by the National Engineering Laboratory for  ... 
arXiv:1911.05277v1 fatcat:sw7nkghsi5hj3bpl5bhhlyq3j4

Large-Scale Place Recognition Based on Camera-LiDAR Fused Descriptor

Shaorong Xie, Chao Pan, Yaxin Peng, Ke Liu, Shihui Ying
2020 Sensors  
captures both the robust representation of the image and 3D point cloud, and (3) learning a proper metric to describe the similarity of our fused global feature.  ...  Our contribution can be summarized as: (1) applying the trimmed strategy in the point cloud global feature aggregation to improve the recognition performance, (2) building a compact fusion framework which  ...  Global Descriptor of 3D Point Cloud via Netvlad VLAD [20] , as mentioned before, is a classical hand-crafted feature extractor method which can aggregate these local features into a single vector representation  ... 
doi:10.3390/s20102870 pmid:32438550 fatcat:xeqtt3ramrcr3abpylkpb3hk7u

Learning Dense Features for Point Cloud Registration Using a Graph Attention Network

Quoc-Vinh Lai-Dang, Sarvar Hussain Nengroo, Hojun Jin
2022 Applied Sciences  
Point cloud registration is a fundamental task in many applications such as localization, mapping, tracking, and reconstruction.  ...  In this paper, we introduce a framework that efficiently and economically extracts dense features using a graph attention network for point cloud matching and registration (DFGAT).  ...  Acknowledgments: This work was supported by an Institute for Information Communications Technology Promotion (IITP) grant funded by the Korean government (MSIT) (No. 2020-0-00440, Development of Artificial  ... 
doi:10.3390/app12147023 fatcat:m7whf37pcrd6rdwfkc4clzzece
« Previous Showing results 1 — 15 out of 6,732 results