Filters








5,013 Hits in 3.0 sec

Point-Voxel CNN for Efficient 3D Deep Learning [article]

Zhijian Liu, Haotian Tang, Yujun Lin, Song Han
2019 arXiv   pre-print
We present Point-Voxel CNN (PVCNN) for efficient, fast 3D deep learning. Previous work processes 3D data using either voxel-based or point-based NN models.  ...  In this paper, we propose PVCNN that represents the 3D input data in points to reduce the memory consumption, while performing the convolutions in voxels to reduce the irregular, sparse data access and  ...  We thank AWS Machine Learning Research Awards for providing the computation resource. We thank NVIDIA for donating Jetson AGX Xavier.  ... 
arXiv:1907.03739v2 fatcat:fsdnj23jcvfu5pkfvczoqio5xe

Multi Voxel-Point Neurons Convolution (MVPConv) for Fast and Accurate 3D Deep Learning [article]

Wei Zhou, Xin Cao, Xiaodan Zhang, Xingxing Hao, Dekui Wang, Ying He
2021 arXiv   pre-print
We present a new convolutional neural network, called Multi Voxel-Point Neurons Convolution (MVPConv), for fast and accurate 3D deep learning.  ...  Moreover, most of the existing 3D deep learning frameworks aim at solving one specific task, and only a few of them can handle a variety of tasks.  ...  Voxel-based Module of Initializing Neuron 3D CNN on voxel grid is a popular selection for state-ofthe-art 3D deep learning researches.  ... 
arXiv:2104.14834v1 fatcat:dxk5olkxbzfczedtcyysqadwny

Multi Point-Voxel Convolution (MPVConv) for Deep Learning on Point Clouds [article]

Wei Zhou, Xin Cao, Xiaodan Zhang, Xingxing Hao, Dekui Wang, Ying He
2021 arXiv   pre-print
The existing 3D deep learning methods adopt either individual point-based features or local-neighboring voxel-based features, and demonstrate great potential for processing 3D data.  ...  Motivated by the success of recent point-voxel representation, such as PVCNN, we propose a new convolutional neural network, called Multi Point-Voxel Convolution (MPVConv), for deep learning on point clouds  ...  MULTI POINT-VOXEL CONVOLUTION State-of-the-art 3D deep learning methods are based on either voxel-based CNN methods or point-based network.  ... 
arXiv:2107.13152v1 fatcat:6ioom3z3yzdkznu6icxdal43sy

MVF-CNN: Fusion of Multilevel Features for Large-scale Point Cloud Classification

Yong Li, Guofeng Tong, Xingang Li, Liqiang Zhang, Hao Peng
2019 IEEE Access  
This paper proposes a deep learning-based algorithm for large-scale point cloud classification through the fusion of multiscale voxels and features (MVF-CNN).  ...  Then, the voxels are input into the 3D convolutional neural network (3D CNN) with three different scale receptive fields for the feature extraction.  ...  network. (2) MS3_DVS [14] : Classification of Point Cloud Scenes with a Multiscale Voxel Deep Network. (3) DeepNet [10] : Deep neural network based on the 3D CNN for point cloud classification, which  ... 
doi:10.1109/access.2019.2908983 fatcat:dsmgb44ubrc3xgne5xgw52x5gq

Multi-level 3D CNN for Learning Multi-scale Spatial Features [article]

Sambit Ghadai, Xian Lee, Aditya Balu, Soumik Sarkar, Adarsh Krishnamurthy
2019 arXiv   pre-print
Current deep learning approaches learn such features either using structured data representations (voxel grids and octrees) or from unstructured representations (graphs and point clouds).  ...  3D object recognition accuracy can be improved by learning the multi-scale spatial features from 3D spatial geometric representations of objects such as point clouds, 3D models, surfaces, and RGB-D data  ...  The presence of abundant information in spatial data coupled with the large data requirement for efficient training of deep learning algorithms render this task impractical for high-resolution 3D data.  ... 
arXiv:1805.12254v2 fatcat:xfmichbn2jgn7gllu2uyknfkzu

Visual Positioning System Based on 6D Object Pose Estimation Using Mobile Web

Ju-Young Kim, In-Seon Kim, Dai-Yeol Yun, Tae-Won Jung, Soon-Chul Kwon, Kye-Dong Jung
2022 Electronics  
The system is inexpensive because it integrates deep learning and computer vision algorithms and does not require additional infrastructure.  ...  methods using RGB depth or point cloud.  ...  Recently, machine learning and deep learning methods are applied without sensors for location recognition.  ... 
doi:10.3390/electronics11060865 fatcat:a6jmccvlyrcgrg7e7cwlov3rra

Fast Point Voxel Convolution Neural Network with Selective Feature Fusion for Point Cloud Semantic Segmentation [article]

Xu Wang, Yuyan Li, Ye Duan
2021 arXiv   pre-print
In contrast to many current CNNs which increase receptive field by downsampling point cloud, our method directly operates on the entire point sets without sampling and achieves good performances efficiently  ...  We present a novel lightweight convolutional neural network for point cloud analysis.  ...  Related Work Volumentric Representation Some early deep learning approaches transformed point clouds into 3D voxel structure and convolve it with standard 3D kernels.  ... 
arXiv:2109.11614v1 fatcat:lyw4o4qgrrgfhf4n2hiq35i7ky

3D-A-Nets: 3D Deep Dense Descriptor for Volumetric Shapes with Adversarial Networks [article]

Mengwei Ren, Liang Niu, Yi Fang
2017 arXiv   pre-print
present in 3D objects. 3D geometric data are often transformed to 3D Voxel grids with regular format in order to be better fed to a deep neural net architecture.  ...  By addressing the challenges posed by the computational inefficiency of direct application of CNN to 3D volumetric data, 3D-A-Nets can learn high-quality 3D-DSDD which demonstrates superior performance  ...  ), recurrent neural network (RNN) and an adversarial discriminator for the robust 3D-DDSD for volumetric shapes.  ... 
arXiv:1711.10108v1 fatcat:7kctixmsuffq5avgdrfeyw6sr4

A Review on Deep Learning Approaches for 3D Data Representations in Retrieval and Classifications

Abubakar Sulaiman Gezawa, Yan Zhang, Qicong Wang, Lei Yunqi
2020 IEEE Access  
Finally, some possible directions for future researches were suggested. INDEX TERMS 3D data representation, 3D deep learning, 3D models dataset, computer vision, classification, retrieval.  ...  Therefore, it can be concluded that deep learning together with a suitable 3D data representation gives an effective approach for improving the performance of 3D shape analysis.  ...  one of the first deep learning methods to utilize the voxels 3D data representation.  ... 
doi:10.1109/access.2020.2982196 fatcat:jnya5rscynf3zm7efuucqxafri

Learning 3D Mineral Prospectivity from 3D Geological Models Using Convolutional Neural Networks: Application to a Structure-controlled Hydrothermal Gold Deposit [article]

Hao Deng
2022 arXiv   pre-print
By exploiting the learning ability of CNNs, the presented method allows for disentangling complex correlation to the mineralization and thus opens a door to circumvent the tedious work for designing the  ...  Motivated by the ability of convolutional neural networks (CNNs) to learn the intrinsic features, in this paper, we present a novel method that leverages CNNs to learn 3D mineral prospectivity from the  ...  In this paper, we propose a novel deep-learning method for learning 3D mineral prospectivity from 3D models, with the aim of providing valuable reference material for 3D prospecting.  ... 
arXiv:2109.00756v2 fatcat:iry3c7dmmneadcsixcg37zxy4u

A New Rotation-Invariant Deep Network for 3D Object Recognition

Yachi Zhang, Zongqing Lu, Jing-Hao Xue, Qingmin Liao
2019 2019 IEEE International Conference on Multimedia and Expo (ICME)  
When inputs are rotated, most 3D convolutional neural networks (CNNs) will have their performance much dropped, especially for those models with voxelized input of 3D objects.  ...  Inspired by this, we propose a new rotation-invariant deep network to recognize rotated 3D objects.  ...  PointNet [8] is designed for deep learning on raw 3D point sets. This network can address the invariance to permutations and transformations of the input points.  ... 
doi:10.1109/icme.2019.00277 dblp:conf/icmcs/ZhangLXL19 fatcat:oz7jch5oojdunhmjidx2rmfdf4

Deep Multi-level Feature Learning on Point Sets for 3D Object Recognition

Yang Xiao, Yanxin Ma, Qianlan Huang, Jun Zhang
2018 DEStech Transactions on Computer Science and Engineering  
In recent years, deep learning has become an important method on point cloud for 3D object recognition. PointNet is the first neural network which could directly consume point cloud as input.  ...  The proposed method achieves a higher accuracy on 3D object recognition with 89.4%.  ...  For most of 3D CNNs, the input of the model should be voxels, occupancy grids.  ... 
doi:10.12783/dtcse/csse2018/24500 fatcat:cqbmd7mqcnbsvhagapiw6ehjmm

A Survey on Deep Geometry Learning: From a Representation Perspective [article]

Yun-Peng Xiao, Yu-Kun Lai, Fang-Lue Zhang, Chunpeng Li, Lin Gao
2020 arXiv   pre-print
Researchers have now achieved great success on dealing with 2D images using deep learning. In recent years, 3D computer vision and Geometry Deep Learning gain more and more attention.  ...  Unlike 2D images, which can be uniformly represented by regular grids of pixels, 3D shapes have various representations, such as depth and multi-view images, voxel-based representation, point-based representation  ...  One of the first models to generate point clouds by deep learning came out in 2017 [21] . They designed a neural network to learn a point sampler based on 3D shape point distribution.  ... 
arXiv:2002.07995v2 fatcat:pustwlu5freypnccfrculkqvei

Deep Learned Full-3D Object Completion from Single View [article]

Dario Rethage, Federico Tombari, Felix Achilles, Nassir Navab
2018 arXiv   pre-print
This writing proposes a new approach to 3D reconstruction and scene understanding, which implicitly learns 3D geometry from depth maps pairing a deep convolutional neural network architecture with an auto-encoder  ...  A data set of synthetic depth views and voxelized 3D representations is built based on ModelNet, a large-scale collection of CAD models, to train networks.  ...  Figure 1 . 1 CNN + Decompressor Architecture elizations and is stacked on the end of the CNN for finetuning.  ... 
arXiv:1808.06843v1 fatcat:sy3irktp55czxglcvn3y4mxrba

A survey on deep geometry learning: From a representation perspective

Yun-Peng Xiao, Yu-Kun Lai, Fang-Lue Zhang, Chunpeng Li, Lin Gao
2020 Computational Visual Media  
Researchers have achieved great success in dealing with 2D images using deep learning. In recent years, 3D computer vision and geometry deep learning have gained ever more attention.  ...  Therefore, in this survey, we review recent developments in deep learning for 3D geometry from a representation perspective, summarizing the advantages and disadvantages of different representations for  ...  One of the first models to generate point clouds by deep learning came out in 2017 [20] . The authors designed a neural network to learn a point sampler based on a 3D point distribution.  ... 
doi:10.1007/s41095-020-0174-8 fatcat:kpoynaixq5esbek63bovybisfa
« Previous Showing results 1 — 15 out of 5,013 results