Filters








106,521 Hits in 5.3 sec

NeRFusion: Fusing Radiance Fields for Large-Scale Scene Reconstruction [article]

Xiaoshuai Zhang, Sai Bi, Kalyan Sunkavalli, Hao Su, Zexiang Xu
2022 arXiv   pre-print
We process the input image sequence to predict per-frame local radiance fields via direct network inference.  ...  These are then fused using a novel recurrent neural network that incrementally reconstructs a global, sparse scene representation in real-time at 22 fps.  ...  In contrast, our network can fuse per-view local reconstruction into a global volume from an arbitrary number of images, leading to highly efficient large-scale scene reconstruction and rendering.  ... 
arXiv:2203.11283v1 fatcat:eldhujgl5fgkbg2ujkjdjjwg2m

Adaptive Local Neighborhood-based Neural Networks for MR Image Reconstruction from Undersampled Data [article]

Shijun Liang, Anish Lahiri, Saiprasad Ravishankar
2022 arXiv   pre-print
Recent works have shown significant promise for reconstructing MR images from sparsely sampled k-space data using deep learning.  ...  Our results demonstrate that our proposed locally-trained method produces higher-quality reconstructions compared to models trained globally on larger datasets.  ...  The approach efficiently learns reconstruction networks from small clusters in flexible training sets and directly at reconstruction time.  ... 
arXiv:2206.00775v1 fatcat:klj7c7766zf3toqigybakz35y4

Deep Local Shapes: Learning Local SDF Priors for Detailed 3D Reconstruction [article]

Rohan Chabra, Jan Eric Lenssen, Eddy Ilg, Tanner Schmidt, Julian Straub, Steven Lovegrove, Richard Newcombe
2020 arXiv   pre-print
This decomposition of scenes into local shapes simplifies the prior distribution that the network must learn, and also enables efficient inference.  ...  DeepLS replaces the dense volumetric signed distance function (SDF) representation used in traditional surface reconstruction systems with a set of locally learned continuous SDFs defined by a neural network  ...  Instead it is more efficient and flexible to encode the space of smaller local shapes and to compose the global shape from an adaptable amount of local codes.  ... 
arXiv:2003.10983v3 fatcat:tesdc5ua6fhppck7daru4ncuoe

Global-Local Face Upsampling Network [article]

Oncel Tuzel, Yuichi Taguchi, John R. Hershey
2016 arXiv   pre-print
In our deep network architecture the global and local constraints that define a face can be efficiently modeled and learned end-to-end using training data.  ...  Conceptually our network design can be partitioned into two sub-networks: the first one implements the holistic face reconstruction according to global constraints, and the second one enhances face-specific  ...  The key element of our algorithm is a deep learning architecture that jointly learns global and local constraints of the high resolution faces.  ... 
arXiv:1603.07235v2 fatcat:tplu5zpvw5bp3efgewjabwby2a

An Approach to 3D Face Reconstruction through Local Deep Feature Alignment

Jian Zhang, Chaoyang Zhu
2018 IET Computer Vision  
, and estimate an explicit mapping from the 2D features to their 3D counterparts for each local neighbourhood, then the authors learn a feed-forward deep neural network for each neighbourhood whose parameters  ...  Here, the authors propose an end-to-end method based on deep learning to reconstruct three-dimensional (3D) face models from given face images.  ...  local features to its global counterparts and the explicit mappings from 3D global features to its local counterparts.  ... 
doi:10.1049/iet-cvi.2018.5151 fatcat:w6asbmf7wzeudgomzvybckmx4e

Unsupervised Feature Learning of Human Actions as Trajectories in Pose Embedding Manifold [article]

Jogendra Nath Kundu, Maharshi Gor, Phani Krishna Uppala, R. Venkatesh Babu
2018 arXiv   pre-print
A hierarchical feature fusion technique is also investigated for simultaneous modeling of local skeleton joints along with global pose variations.  ...  In contrast to end-to-end framework explored by previous works, we disentangle the task of individual pose representation learning from the task of learning actions as a trajectory in pose embedding space  ...  Acknowledgements This work was supported by a CSIR Fellowship (Jogendra), and a project grant from Robert Bosch Centre for Cyber-Physical Systems, IISc.  ... 
arXiv:1812.02592v1 fatcat:37jnz4444faudnvaiao5d6sjym

Image Super-Resolution via Simplified Dense Network with Non-degenerate Layers

Zhimin Tang, Shaohui Li, Linkai Luo, Min Fu, Hong Peng, Qifeng Zhou
2019 IEEE Access  
INDEX TERMS Convolutional neural network, deep learning, image restoration, image super-resolution.  ...  In this paper, we present an efficient method for single image super-resolution based on the dense network.  ...  The efficiency of our method benefits from the efficient global structure, local structure, direct reconstruct module and the perceptual prior.  ... 
doi:10.1109/access.2019.2898846 fatcat:4idhniust5abzlgrw3bd6fxrl4

Network Embedding via Community Based Variational Autoencoder

Wei Shi, Ling Huang, Chang-Dong Wang, Juan-Hui Li, Yong Tang, Chengzhou Fu
2019 IEEE Access  
Second, deep learning techniques can not only integrate and preserve the information from both local and global views efficiently but also strengthen the robustness of vertex representations.  ...  First, community information reveals an implicit relationship between vertices from a global view, which can be a supplement to local information and help to improve the embedding quality.  ...  Firstly, deep learning techniques can not only integrate the information from both local and global views efficiently, but also strengthen the robustness of representations.  ... 
doi:10.1109/access.2019.2900662 fatcat:vhsz7puq5vc7rb2xkbe4vei7gm

Learning Deep Implicit Functions for 3D Shapes with Dynamic Code Clouds [article]

Tianyang Li, Xin Wen, Yu-Shen Liu, Hua Su, Zhizhong Han
2022 arXiv   pre-print
In contrast to previous methods, our DCC-DIF represents 3D shapes more efficiently with a small amount of local codes, and improves the reconstruction quality.  ...  To capture geometry details, current methods usually learn DIF using local latent codes, which discretize the space into a regular 3D grid (or octree) and store local codes in grid points (or octree nodes  ...  Different from explicit 3D representations, DIF can be stored compactly and learn shape priors by the network.  ... 
arXiv:2203.14048v2 fatcat:qtiv23sjnreafjvbquxqszptau

Similarity Hashing and Learning for Tracks Reconstruction

Sabrina Amrouche, Tobias Golling, Moritz Kiehn, Andreas Salzburger
2019 Zenodo  
A neural network selects valid combinations in the buckets and builds up full trajectories by connected components search independently of global positions of the hits and detector geometry.  ...  We propose a framework for Similarity Hashing and Learning for Track Reconstruction (SHLTR) where multiple small regions of the detector are reconstructed in parallel with minimal fake rate.  ...  (similarity) Learning 12 Metric LearningLearn a projection that improves tracks separability ○ Local Fischer Discriminant Analysis ( LFDA 1 ) ○ Solved as a generalized eigenvalue problem  ... 
doi:10.5281/zenodo.3599393 fatcat:bjezwy6edfdctosvk7lpwzzhbi

Do not Lose the Details: Reinforced Representation Learning for High Performance Visual Tracking

Qiang Wang, Mengdan Zhang, Junliang Xing, Jin Gao, Weiming Hu, Steve Maybank
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
The correlation filter in this layer is updated online efficiently without network fine-tuning.  ...  In particular, a fully convolutional encoder-decoder network is designed to reconstruct the original visual features from the semantic projections to preserve all the geometric information.  ...  This approach is more suitable for efficient tracking and accurate localization. Deep learning based tracking.  ... 
doi:10.24963/ijcai.2018/137 dblp:conf/ijcai/WangZXGHM18 fatcat:vuixvii3hfdw5pp5snso2kxaom

Ray-ONet: Efficient 3D Reconstruction From A Single RGB Image [article]

Wenjing Bian and Zirui Wang and Kejie Li and Victor Adrian Prisacariu
2021 arXiv   pre-print
We propose Ray-ONet to reconstruct detailed 3D models from monocular images efficiently.  ...  Networks (ONet), while reducing the network inference complexity to O(N^2).  ...  network [16] , with different inputs and output dimensions. The inputs are the global feature (z, s) and the local feature output from J .  ... 
arXiv:2107.01899v2 fatcat:yd6tjvaoe5aghmzqttagwzn7ke

SSRNet: Scalable 3D Surface Reconstruction Network [article]

Zhenxing Mi, Yiming Luo, Wenbing Tao
2020 arXiv   pre-print
Existing learning-based surface reconstruction methods from point clouds are still facing challenges in terms of scalability and preservation of details on large-scale point clouds.  ...  The proposed SSRNet constructs local geometry-aware features for octree vertices and designs a scalable reconstruction pipeline, which not only greatly enhances the predication accuracy of the relative  ...  The tangent convolution network [28] learns local features from neighbor points for semantic segmentation of 3D point clouds.  ... 
arXiv:1911.07401v2 fatcat:5onwutntg5dqvlw4roqwfnp7cu

SSRNet: Scalable 3D Surface Reconstruction Network

Zhenxing Mi, Yiming Luo, Wenbing Tao
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Existing learning-based surface reconstruction methods from point clouds are still facing challenges in terms of scalability and preservation of details on large-scale point clouds.  ...  The proposed SSRNet constructs local geometry-aware features for octree vertices and designs a scalable reconstruction pipeline, which not only greatly enhances the predication accuracy of the relative  ...  The tangent convolution network [28] learns local features from neighbor points for semantic segmentation of 3D point clouds.  ... 
doi:10.1109/cvpr42600.2020.00105 dblp:conf/cvpr/MiLT20 fatcat:vfgtwrnqubg4th237nqb5ihxpe

Contextual Feature Constrained Semantic Face Completion with Paired Discriminator

Xiuhong Yang, Peng Xu, Yi Xue, Haiyan Jin
2021 IEEE Access  
Our network is composed of an encoder-decoder generator and a local and global (paired) adversarial discriminator.  ...  reconstruction losses.  ...  In order to thoroughly learn the deep features of the image, we try to add local and global feature reconstruction loss on the basis of dual-discriminator network to enhance the authenticity and global  ... 
doi:10.1109/access.2021.3065661 fatcat:fr6pl44bxvhxvc3o4yqkem7paq
« Previous Showing results 1 — 15 out of 106,521 results