全部 |
  • 全部
  • 题名
  • 关键词
  • NSTL主题词
  • 摘要
  • 会议名称
  • 论文-出处
  • 论文-作者
  • 论文-机构
  • 论文-DOI
  • 会议-出版者
  • 会议-出版地
  • 会议-主编
  • 会议-主办单位
  • 会议-举办地
  • ISSN
  • EISSN
  • ISBN
  • EISBN
检索 搜索会议录 二次检索 AI检索
展示更多
外文文献 中文文献
筛选条件:

1. Do Text-Free Diffusion Models Learn Discriminative Visual Representations? NSTL国家科技图书文献中心

Soumik Mukhopadhyay |  Matthew Gwilliam... -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 253~272 - 共20页

摘要:Diffusion models have proven to be state-of | -the-art methods for generative tasks. These models |  involve training a U-Net to iteratively predict and |  remove noise, and the resulting model can synthesize |  high-fidelity, diverse, novel images. However, text
关键词: Diffusion models |  Representation learning |  Self-Supervised learning

2. VF-NeRF: Viewshed Fields for Rigid NeRF Registration NSTL国家科技图书文献中心

Leo Segre |  Shai Avidan -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 164~181 - 共18页

摘要:3D scene registration is a fundamental problem |  in computer vision that seeks the best 6-DoF |  alignment between two scenes. This problem was extensively |  investigated in the case of point clouds and meshes, but |  there has been relatively limited work regarding
关键词: Neural radiance fields |  3D registration |  Normalizing-flows

3. Distributed Active Client Selection With Noisy Clients Using Model Association Scores NSTL国家科技图书文献中心

Kwang In Kim -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 75~92 - 共18页

摘要:Active client selection (ACS) strategically |  identifies clients for model updates during each training |  round of federated learning. In scenarios with limited |  communication resources, ACS emerges as a superior alternative |  to random client selection, significantly improving
关键词: Active client selection |  Federated learning |  Label noise

4. AttentionHand: Text-Driven Controllable Hand Image Generation for 3D Hand Reconstruction in the Wild NSTL国家科技图书文献中心

Junho Park |  Kyeongbo Kong... -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 329~345 - 共17页

摘要:Recently, there has been a significant amount |  of research conducted on 3D hand reconstruction to |  use various forms of human-computer interaction | . However, 3D hand reconstruction in the wild is |  challenging due to extreme lack of in-the-wild 3D hand
关键词: 3D hand mesh reconstruction |  Text-to-Image generation

5. Pseudo-Keypoint RKHS Learning for Self-supervised 6DoF Pose Estimation NSTL国家科技图书文献中心

Yangzheng Wu |  Michael Greenspan -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 37~56 - 共20页

摘要:We address the simulation-to-real domain gap |  in six degree-of-freedom pose estimation (6DoF PE | ), and propose a novel self-supervised keypoint voting | -based 6DoF PE framework, effectively narrowing this |  gap using a learnable kernel in RKHS. We formulate
关键词: Pose estimation |  Self-supervision |  Domain adaptation |  Keypoint estimation

6. Match-Stereo-Videos: Bidirectional Alignment for Consistent Dynamic Stereo Matching NSTL国家科技图书文献中心

Junpeng Jing |  Ye Mao... -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 415~432 - 共18页

摘要:Dynamic stereo matching is the task of |  estimating consistent disparities from stereo videos with |  dynamic objects. Recent learning-based methods |  prioritize optimal performance on a single stereo pair | , resulting in temporal inconsistencies. Existing video
关键词: Dynamic stereo matching |  Bidirectional alignment

7. Sur~2f: A Hybrid Representation for High-Quality and Efficient Surface Reconstruction from Multi-view Images NSTL国家科技图书文献中心

Zhangjin Huang |  Zhihao Liang... -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 1~18 - 共18页

摘要:Multi-view surface reconstruction is an ill | -posed, inverse problem in 3D vision research. It |  involves modeling the geometry and appearance with |  appropriate surface representations. Most of the existing |  methods rely either on explicit meshes, using surface
关键词: Multi-view reconstruction |  3D representation |  Neural volume rendering |  Physics-based rendering |  Differentiable rendering

8. HO-Gaussian: Hybrid Optimization of 3D Gaussian Splatting for Urban Scenes NSTL国家科技图书文献中心

Zhuopeng Li |  Yilin Zhang... -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 19~36 - 共18页

摘要:The rapid growth of 3D Gaussian Splatting |  (3DGS) has revolutionized neural rendering, enabling |  real-time production of high-quality renderings | . However, the previous 3DGS-based methods have |  limitations in urban scenes due to reliance on initial
关键词: Novel view synthesis |  Urban scenes |  Gaussian splatting

9. Trainable Highly-Expressive Activation Functions NSTL国家科技图书文献中心

Irit Chelly |  Shahaf E. Finder... -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 200~217 - 共18页

摘要:Nonlinear activation functions are pivotal to |  the success of deep neural nets, and choosing the |  appropriate activation function can significantly affect |  their performance. Most networks use fixed activation |  functions (e.g., ReLU, GELU, etc.), and this choice might
关键词: Trainable activation functions |  Diffeomorphisms |  Deep learning

10. GarmentCodeData: A Dataset of 3D Made-to-Measure Garments with Sewing Patterns NSTL国家科技图书文献中心

Maria Korosteleva |  Timur Levent Kesdoga...... -  《Computer Vision - ECCV 2024,Part LX》 -  European Conference on Computer Vision - 2025, - 110~127 - 共18页

摘要:Recent research interest in learning-based |  processing of garments, from virtual fitting to generation |  and reconstruction, stumbles on a scarcity of high | -quality public data in the domain. We contribute to |  resolving this need by presenting the first large-scale
关键词: Garments data |  Dataset generation |  Cloth reconstruction
检索条件出处:Computer Vision - ECCV 2024,Part LX

NSTL主题词

  • NSTL学科导航