检索结果-内蒙古大学图书馆

Efficient Reflectance Capture With a Deep Gated Mixture-of-Experts

IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS 2024年第7期30卷 4246-4256页

作者： Ma, Xiaohe Yu, Yaxin Wu, Hongzhi Zhou, Kun Zhejiang Univ State Key Lab CAD & CG Hangzhou 310058 Peoples R China ZJU FaceUn Joint Lab Intelligent G Hangzhou 310058 Peoples R China

We present a novel framework to efficiently acquire anisotropic reflectance in a pixel-independent fashion, using a deep gated mixture-of-experts. While existing work employs a unified network to handle all possible input, our network automatically learns to condition on the input for enhanced reconstruction. We train a gating module that takes photometric measurements as input and selects one out of a number of specialized decoders for reflectance reconstruction, essentially trading generality for quality. A common pre-trained latent-transform module is also appended to each decoder, to offset the burden of the increased number of decoders. In addition, the illumination conditions during acquisition can be jointly optimized. The effectiveness of our framework is validated on a wide variety of challenging near-planar samples with a lightstage. Compared with the state-of-the-art technique, our quality is improved with the same number of input images, and our input image number can be reduced to about 1/3 for equal-quality results. We further generalize the framework to enhance a state-of-the-art technique on non-planar reflectance scanning.

关键词： Lighting Image reconstruction Decoding Optimization Light emitting diodes Cameras Neural networks Anisotropic reflectance computational illumination SVBRDF

来源：评论

学校读者我要写书评

暂无评论

SNPF: Sensitiveness-Based Network Pruning Framework for Efficient Edge Computing

引用

IEEE INTERNET OF THINGS JOURNAL 2024年第4期11卷 6972-6991页

作者： Lu, Yiheng Guan, Ziyu Zhao, Wei Gong, Maoguo Wang, Wenxiao Sheng, Kai Xidian Univ Minist Educ Key Lab Collaborat Intelligence Syst Xian 710071 Peoples R China Zhejiang Univ State Key Lab CAD&CG Hangzhou Peoples R China

Convolutional neural networks (CNNs) are used comprehensively in the field of the Internet of Things (IoTs), such as mobile phones, surveillance, and satellite. However, the deployment of CNNs is difficult because the structure of hand-designed networks is complicated. Therefore, we propose a sensitiveness-based network pruning framework (SNPF) to reduce the size of original networks to save computation resources. SNPF will evaluate the importance of each convolutional layer by the reconstruction of inference accuracy when we add extra noise to the original model, and then remove filters in terms of the degree of sensitiveness for each layer. Compared with previous weight-norm-based pruning methods, such as " C-1-norm," "BatchNorm-Pruning," and "Taylor-Pruning," SNPF is robust to the update of parameters, which can avoid the inconsistency of evaluation for filters if the parameters of the pretrained model are not fully optimized. Namely, SNPF, can prune the network at the early training stage to save computation resources. We test our method on three prevalent models of VGG-16, ResNet-18, ResNet-50 and a customized Conv-4 with 4 convolutional layers. They are then tested on CIFAR-10, CIFAR-100, ImageNet, and MNIST, respectively. Impressively, we observe that even when the VGG-16 is only trained with 50 epochs, we can get the same evaluation of layer importance as the results when the model is fully trained. Additionally, we can also achieve comparable pruning results to previous weight-oriented methods on the other three models.

关键词： Computational modeling Information filters Training Neurons Internet of Things Biological neural networks Surveillance Convolutional neural network network pruning sensitiveness

来源：评论

学校读者我要写书评

暂无评论

NeLT: Object-Oriented Neural Light Transfer

引用

ACM TRANSACTIONS ON GRAPHICS 2023年第5期42卷 1-16页

作者： Zheng, Chuankun Huo, Yuchi Mo, Shaohua Zhong, Zhihua Wu, Zhizhen Hua, Wei Wang, Rui Bao, Hujun Zhejiang Univ State Key Lab CAD&CG Hangzhou Peoples R China Zhejiang Lab Hangzhou Peoples R China

This article presents object-oriented neural light transfer (NeLT), a novel neural representation of the dynamic light transportation between an object and the environment. Our method disentangles the global illumination of a scene into individual objects' light transportation represented via neural networks, then composes them explicitly. It therefore enables flexible rendering with dynamic lighting, cameras, materials, and objects. Our rendering features various important global illumination effects, such as diffuse illumination, glossy illumination, dynamic shadowing, and indirect illumination, which completes the capability of existing neural object representation. Experiments show that NeLT does not require path tracing or shading results as input but achieves rendering quality comparable to state-of-the-art rendering frameworks, including the recent deep learning based denoisers.

关键词： Neural rendering radiance transfer

来源：评论

学校读者我要写书评

暂无评论

Differentiable Physical Optics Method for 3-D Reconstruction of Dielectric Objects

Differentiable Physical Optics Method for 3-D Reconstruction...

引用

IEEE International Symposium on Antennas and Propagation / INC/USNC-URSI Radio Science Meeting (AP-S/INC-USNC-URSI)

作者： Yang, Lizhen Lin, Hai Zhejiang Univ State Key Lab CAD&CG Hangzhou 310058 Peoples R China

ISBN: (纸本)9798350369908;9798350369915

This paper presents a novel machine learning framework for 3-D reconstruction of dielectric targets from scattered electromagnetic (EM) field data. Our framework comprises two pivotal components: the shape representation part, which utilizes neural implicit surfaces as the continuous geometric representation of the reconstructed model, and the EM parameters storage part, which employs a fully-connected neural network to store the EM properties of the reconstructed model. Our algorithm takes the incident wave configuration as input and outputs the corresponding scattered field computed by the differentiable physical optics method (DPO). Subsequently, we optimize the geometric shape and the EM parameters by minimizing the residual between the output and the measured scattered field. Our numerical experiments demonstrate promising results in 3-D reconstruction tasks.

关键词： 3D reconstruction

来源：评论

学校读者我要写书评

暂无评论

A Progressive Sampling Method for 3-D Reconstruction of DPO-SDF

A Progressive Sampling Method for 3-D Reconstruction of DPO-...

引用

IEEE International Symposium on Antennas and Propagation / INC/USNC-URSI Radio Science Meeting (AP-S/INC-USNC-URSI)

作者： Zhang, Ruoming Shen, Yuhao Shen, Qianyan Yang, Lizhen Lin, Hai Zhejiang Univ State Key Lab CAD&CG Hangzhou 310058 Peoples R China

ISBN: (纸本)9798350369908;9798350369915

The Differentiable Physical Optics Method with Signed Distance Field (DPO-SDF) facilitates the 3-D reconstruction of perfect electric conductors (PEC) using scattered electric fields, achieving high precision with low computational costs. However, its uniform sampling technique leads to an overuse of memory and increased time consumption due to redundant sampling points. As an alternative, this paper proposes a heterogeneous ruleset sampling method, which includes progressive sampling to rapidly approximate the target boundary through an iterative approach, and uniform sampling in the vicinity of the target to ensure accuracy. Numerical experiments indicate that this method achieves a 46.7% reduction in model training time, and the chamfer distance is reduced by nearly 72.4%, indicating a significant improvement in the reconstructed model precision.

关键词： Physical optics

来源：评论

学校读者我要写书评

暂无评论

Gaussian Surfel Splatting for Live Human Performance Capture

引用

ACM TRANSACTIONS ON GRAPHICS 2024年第6期43卷 1-17页

作者： Dong, Zheng Xu, Ke Gao, Yaoan Bao, Hujun Xu, Weiwei Lau, Rynson w. h. Zhejiang Univ State Key Lab CAD&CG Zhejiang Peoples R China City Univ Hong Kong Hong Kong Peoples R China

High-quality real-time rendering using user-affordable capture rigs is an essential property of human performance capture systems for real-world applications. However, state-of-the-art performance capture methods may not yield satisfactory rendering results under a very sparse (e.g., four) capture setting. Specifically, neural radiance field (NeRF)-based methods and 3D Gaussian Splatting (3DGS)-based methods tend to produce local geometry errors for unseen performers, while occupancy field (PIFu)-based methods often produce unrealistic rendering results. In this paper, we propose a novel generalizable neural approach to reconstruct and render the performers from very sparse RGBD streams in high quality. The core of our method is a novel point-based generalizable human (PGH) representation conditioned on the pixel-aligned RGBD features. The PGH representation learns a surface implicit function for the regression of surface points and a Gaussian implicit function for parameterizing the radiance fields of the regressed surface points with 2D Gaussian surfels, and uses surfel splatting for fast rendering. We learn this hybrid human representation via two novel networks. First, we propose a novel point-regressing network (PRNet) with a depth-guided point cloud initialization (DPI) method to regress an accurate surface point cloud based on the denoised depth information. Second, we propose a novel neural blending-based surfel splatting network (SPNet) to render high-quality geometries and appearances in novel views based on the regressed surface points and high-resolution RGBD features of adjacent views. Our method produces free-view human performance videos of 1K resolution at 12 fps on average. Experiments on two benchmarks show that our method outperforms state-of-the-art human performance capture methods.

关键词： human performance capture point-based generalizable human representation surface implicit function Gaussian implicit function

来源：评论

学校读者我要写书评

暂无评论

Efficient LoFTR: Semi-Dense Local Feature Matching with Sparse-Like Speed

Efficient LoFTR: Semi-Dense Local Feature Matching with Spar...

引用

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Wang, Yifan He, Xingyi Peng, Sida Tan, Dongli Zhou, Xiaowei Zhejiang Univ Affiliated State Key Lab CAD&CG Hangzhou Peoples R China

ISBN: (纸本)9798350353006

We present a novel method for efficiently producing semi-dense matches across images. Previous detector-free matcher LoFTR has shown remarkable matching capability in handling large-viewpoint change and texture-poor scenarios but suffers from low efficiency. We revisit its design choices and derive multiple improvements for both efficiency and accuracy. One key observation is that performing the transformer over the entire feature map is redundant due to shared local information, therefore we propose an aggregated attention mechanism with adaptive token selection for efficiency. Furthermore, we find spatial variance exists in LoFTR's fine correlation module, which is adverse to matching accuracy. A novel two-stage correlation layer is proposed to achieve accurate subpixel correspondences for accuracy improvement. Our efficiency optimized model is similar to 2.5x faster than LoFTR which can even surpass state-of-the-art efficient sparse matching pipeline SuperPoint + LightGlue. Moreover, extensive experiments show that our method can achieve higher accuracy compared with competitive semi-dense matchers, with considerable efficiency benefits. This opens up exciting prospects for large-scale or latency-sensitive applications such as image retrieval and 3D reconstruction. Project page: https://***/efficientloftr/.

关键词： Image matching

来源：评论

学校读者我要写书评

暂无评论

A 4D-Printed Structure With Reversible Deformation for the Soft Crawling Robot

引用

FRONTIERS IN MATERIALS 2022年 9卷

作者： Duan, Guifang Liu, Han Liu, Zhenyu Tan, Jianrong Zhejiang Univ State Key Lab CAD & CG Hangzhou Peoples R China

Reversible deformations of the 4D-printed structures are attractive and promising for various application fields. In this study, the principle of reversible deformations for the bilayer structure consisting of SMP and elastic material is illustrated. By exploring the influence of printing parameters on deformation and resistance, a low-cost reversible bilayer structure with rational resistance distribution is designed to realize reversible deformation. Subsequently, the bilayer structure is employed to design a soft crawling robot with asymmetrical variable friction coefficient feet. By revealing the principle of locomotion by force analysis and deformation process analysis, a wave-like strategy is proposed to actuate the robot. Experiments verify the effectiveness of the designed structures.

关键词： reversible deformation soft crawling robot 4D printing composite structure force and deformation analysis

来源：评论

学校读者我要写书评

暂无评论

Economic Upper Bound Estimation in Hausdorff Distance Computation for Triangle Meshes

引用

COMPUTER GRAPHICS FORUM 2022年第1期41卷 46-56页

作者： Zheng, Yicun Sun, Haoran Liu, Xinguo Bao, Hujun Huang, Jin Zhejiang Univ State Key Lab CAD & CG Hangzhou Peoples R China

The Hausdorff distance is one of the most fundamental metrics for comparing 3D shapes. To compute the Hausdorff distance efficiently from a triangular mesh A to another triangular mesh B, one needs to cull the unnecessary triangles on A quickly. These triangles have no chance to improve the Hausdorff distance estimation, that is the parts with local upper bound smaller than the global lower bound. The local upper bound estimation should be tight, use fast distance computation, and involve a small number of triangles in B during the reduction phase for efficiency. In this paper, we propose to use point-triangle distance, and only involve at most four triangles in B in the reduction phase. Comparing with the state-of-the-art proposed by Tang et al. in 2009, which uses more costly triangle-triangle distance and may involve a large number of triangles in reduction phase, our local upper bound estimation is faster, and with only a small impact on the tightness of the bound on error estimation. Such a more economic strategy boosts the overall performance significantly. Experiments on the Thingi10K dataset show that our method can achieve several (even over 20) times speedup on average. On a few models with different placements and resolutions, we show that close placement and large difference in resolution bring big challenges to Hausdorff distance computation, and explain why our method can achieve more significant speedup on challenging cases.

关键词： computational geometry modeling

来源：评论

学校读者我要写书评

暂无评论

X-View: Non-Egocentric Multi-View 3D Object Detector

引用

IEEE TRANSACTIONS ON IMAGE PROCESSING 2023年 32卷 1488-1497页

作者： Xie, Liang Xu, Guodong Cai, Deng He, Xiaofei Zhejiang Univ State Key Lab Comp Aided Design CAD & Comp CG G Hangzhou 310027 Peoples R China

3D object detection algorithms for autonomous driving reason about 3D obstacles either from 3D birds-eye view or perspective view or both. Recent works attempt to improve the detection performance via mining and fusing from multiple egocentric views. Although the egocentric perspective view alleviates some weaknesses of the birds-eye view, the sectored grid partition becomes so coarse in the distance that the targets and surrounding context mix together, which makes the features less discriminative. In this paper, we generalize the research on 3D multi-view learning and propose a novel multi-view-based 3D detection method, named X-view, to overcome the drawbacks of the multi-view methods. Specifically, X-view breaks through the traditional limitation about the perspective view whose original point must be consistent with the 3D Cartesian coordinate. X-view is designed as a general paradigm that can be applied on almost any 3D detectors based on LiDAR with only little increment of running time, no matter it is voxel/grid-based or raw-point-based. We conduct experiments on KITTI and NuScenes datasets to demonstrate the robustness and effectiveness of our proposed X-view. The results show that X-view obtains consistent improvements when combined with mainstream state-of-the-art 3D methods.

关键词： Three-dimensional displays Feature extraction Detectors Laser radar Point cloud compression Interpolation Fuses 3D object detection multi-view fusion autonomous driving

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：