- 综合排序
- 最热优先
- 最新优先
- 来自专栏计算机视觉理论及其实现
StyleSwin: Transformer-based GAN for High-resolution Image Generation
尽管Transformer在广泛的视觉任务中取得了诱人的成功，但在高分辨率图像生成建模方面，Transformer还没有表现出与ConvNets同等的能力。在本文中，我们试图探索使用Transformer来构建用于高分辨率图像合成的生成对抗性网络。为此，我们认为局部注意力对于在计算效率和建模能力之间取得平衡至关重要。因此，所提出的生成器在基于风格的架构中采用了Swin Transformer。为了实现更大的感受野，我们提出了双重关注，它同时利用了局部窗口和偏移窗口的上下文，从而提高了生成质量。此外，我们表明，提供基于窗口的Transformer中丢失的绝对位置的知识极大地有利于生成质量。所提出的StyleSwan可扩展到高分辨率，粗糙的几何结构和精细的结构都得益于Transformer的强大表现力。然而，在高分辨率合成期间会出现块伪影，因为以块方式执行局部关注可能会破坏空间相干性。为了解决这个问题，我们实证研究了各种解决方案，其中我们发现使用小波鉴别器来检查频谱差异可以有效地抑制伪影。大量实验表明，它优于现有的基于Transformer的GANs，尤其是在高分辨率（例如1024×1024）方面。StyleWin在没有复杂训练策略的情况下，在CelebA HQ 1024上优于StyleGAN，在FFHQ-1024上实现了同等性能，证明了使用Transformer生成高分辨率图像的前景。
1.3K20编辑于 2023-10-07 - 来自专栏数据分析与挖掘
(HRNet)：Deep High-Resolution Representation Learning for Visual Recognition相关论文
github：https://github.com/HRNet 论文地址：https://arxiv.org/pdf/1908.07919 相关论文： 1.Deep High-Resolution Representation Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation 4.Multi-Stage HRNet: Multiple Stage High-Resolution
67240发布于 2020-08-26 - 来自专栏YOLO大作战
《Towards High-Resolution Industrial Image Anomaly Detection》（迈向高分辨率工业图像异常检测）
本文摘要：该论文《Towards High-Resolution Industrial Image Anomaly Detection》（迈向高分辨率工业图像异常检测）具有以下重要意义：1. ，1）YOLOv5、v7、v8、v9、v10、11优化创新，轻松涨点和模型轻量化；2）目标检测、语义分割、OCR、分类等技术孵化，赋能智能制造，工业项目落地经验丰富；1.原理介绍论文：Towards High-Resolution
61810编辑于 2026-01-10 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
GEE数据集——全球SWOT 河流数据库（SWORD）河流水面高程、宽度和坡度测量数据集
Reach ID, N: Node ID within reach, T: Type) none node_length Length of the node measured along the high-resolution Average water surface elevation of the node meters wse_var Variance of water surface elevation along the high-resolution for each node meters^2 width Average width of the node meters width_var Variance of width along the high-resolution Pfafstetter basin codes, R: Reach ID, T: Type) none reach_length Length of the reach measured along the high-resolution for each reach meters^2 width Average width of the reach meters width_var Variance of width along the high-resolution
1.2K10编辑于 2024-06-12 - 来自专栏气象学家
ECMWF实时开放数据官方说明、下载
(HRES) forecasts Examples using wget for products based on the Wave Model high-resolution (HRES-WAM) Values are: oper - high-resolution forecast, atmospheric fields enfo - ensemble forecast, atmospheric fields waef - ensemble forecast, ocean wave fields, scda - short cut-off high-resolution forecast, atmospheric fields (also known as "high-frequency products") scwv - short cut-off high-resolution forecast Examples using wget for products based on the Atmospheric Model high-resolution (HRES) forecasts In the
4.5K51编辑于 2022-03-29 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine——世界人口数据集描述了2010年、2015年和其他年份居住在每个网格单元的估计人数。
WorldPop Project Population Data: Estimated Residential Population per 100x100m Grid Square [deprecated] High-resolution Tatem, 2015, High-resolution gridded population datasets for Latin America and the Caribbean in 2010, 2015, and 2020, Scientific Data, [doi:10.1038/sdata.2015.45] (High-resolution gridded population datasets
46810编辑于 2024-02-02 - 来自专栏AI研习社
CVPR2019 | 15篇论文速递（涵盖目标检测、语义分割和姿态估计等方向）
姿态估计 [1] CVPR 2019 Pose estimation文章，目前SOTA,已经开源 论文题目：Deep High-Resolution Representation Learning for Most existing methods recover high-resolution representations from low-resolution representations produced Instead, our proposed network maintains high-resolution representations through the whole process. 多视几何 [14] CVPR2019 多视几何新文 论文题目：Recurrent MVSNet for High-resolution Multi-view Stereo Depth Inference This reduces dramatically the memory consumption and makes high-resolution reconstruction feasible.
1.1K30发布于 2019-05-15 - 来自专栏机器学习、深度学习
语义分割
RefineNet: Multi-Path Refinement Networks for High-Resolution Semantic Segmentation CVPR2017 https 再就是 a large number of high-dimensional and high-resolution feature maps 导致需要 较大的 GPU 内存 ，尤其在训练时，这就导致输出结果的尺寸是输入的
1K10发布于 2019-05-26 - 来自专栏DeveWork
iOS / Android 移动设备中的 Touch Icons
-- For iPhone with high-resolution Retina display running iOS ≤ 6: --> <link rel="apple-touch-icon-precomposed -- For iPhone with <em>high-resolution</em> Retina display running iOS ≥ 7: --> <link rel="apple-touch-icon-precomposed -- For iPad with high-resolution Retina display running iOS ≤ 6: --> <link rel="apple-touch-icon-precomposed -- For iPad with <em>high-resolution</em> Retina display running iOS ≥ 7: --> <link rel="apple-touch-icon-precomposed
2.8K60发布于 2018-01-19 - 来自专栏图像处理与模式识别研究所
视频抽帧
Extraction of high-resolution frames from video sequences. Extraction of high-resolution frames from video sequences.
1.6K10编辑于 2022-05-29 - 来自专栏机器之心
7 Papers & Radios | 阿里达摩院获KDD 2022最佳论文；Meta发布110亿参数模型
FederatedScope-GNN: Towards a Unified, Comprehensive and Efficient Package for Federated Graph Learning High-Resolution 论文 2：High-Resolution Image Synthesis with Latent Diffusion Models 作者：Robin Rombach 、 Andreas Blattmann 等 论文地址：https://arxiv.org/pdf/2112.10752.pdf 摘要：来自慕尼黑大学和 Runway 的研究者基于其 CVPR 2022 的论文《High-Resolution 论文 5：High-Resolution Image Synthesis with Latent Diffusion Models 作者：Robin Rombach 等 论文地址：https://arxiv.org 现在，来自慕尼黑大学和 Runway 的研究者基于其 CVPR 2022 的论文《High-Resolution Image Synthesis with Latent Diffusion Models
65410编辑于 2022-08-25 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine（GEE）——全球高分辨率洪泛区（GFPLAIN250m）数据集
GFPLAIN250m, a global high-resolution dataset of Earth’s floodplains | Scientific Data 高程数据由一个快速地理空间工具处理 GFPLAIN250m, a global high-resolution dataset of Earth’s floodplains. Sci.
43310编辑于 2024-02-02 - 来自专栏个人博客
StableDiffusion笔记 - plus studio
High-Resolution Image Synthesis with Latent Diffusion Models. 2022 IEEE/CVF Conference on Computer Vision GitHub - Stability-AI/stablediffusion: High-Resolution image synthesis with latent diffusion models.
39210编辑于 2024-02-28 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine ——全球JRC/GSW1_1/YearlyHistory数据集的观测数据
For more information see the associated journal article: High-resolution mapping of global surface water Belward, High-resolution mapping of global surface water and its long-term changes.
23510编辑于 2024-02-02 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine ——全球JRC/GSW1_2/Metadata数据集的观测数据
For more information see the associated journal article: High-resolution mapping of global surface water Belward, High-resolution mapping of global surface water and its long-term changes.
28710编辑于 2024-02-02 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine ——全球1984年至2015年地表水的位置和时间即地表水数据集的观测数据的元数据
For more information see the associated journal article: High-resolution mapping of global surface water Belward, High-resolution mapping of global surface water and its long-term changes.
32610编辑于 2024-02-02 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine ——全球JRC/GSW1_1/MonthlyRecurrence数据集的观测数据
For more information see the associated journal article: High-resolution mapping of global surface water Belward, High-resolution mapping of global surface water and its long-term changes.
31910编辑于 2024-02-02 - 来自专栏专知
【论文推荐】最新6篇生成式对抗网络（GAN）相关论文—半监督对抗学习、行人再识别、代表性特征、高分辨率深度卷积、自监督、超分辨
High-Resolution Deep Convolutional Generative Adversarial Networks（高分辨率深度卷积生成对抗网络） ---- ---- 作者：Joachim Lyu 摘要：Generative Adversarial Networks (GANs) convergence in a high-resolution setting with a computational convergence of DCGAN (Deep Convolutional Generative Adversarial Networks) and achieve good-looking high-resolution Our experiments show that the generator is able to infer more realistic high-resolution details by using
1.7K60发布于 2018-04-13 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine ——全球JRC/GSW1_1/Metadata数据集的观测数据的元数据
For more information see the associated journal article: High-resolution mapping of global surface water Belward, High-resolution mapping of global surface water and its long-term changes.
31810编辑于 2024-02-02 - 来自专栏GEE数据专栏，GEE学习专栏，GEE错误集等专栏
Google Earth Engine ——全球JRC/GSW1_1/MonthlyHistory数据集的观测数据
For more information see the associated journal article: High-resolution mapping of global surface water Belward, High-resolution mapping of global surface water and its long-term changes.
30810编辑于 2024-02-02
腾讯云开发者
Copyright © 2013 - 2026 Tencent Cloud. All Rights Reserved. 腾讯云 版权所有
深圳市腾讯计算机系统有限公司 ICP备案/许可证号：粤B2-20090059 
粤公网安备44030502008569号
腾讯云计算（北京）有限责任公司 京ICP证150476号 | 京ICP备11018762号