ZeyuXiao @ USTC

Paper reading notes, code sharing platforms


  • Home

  • About

  • Tags48

  • Categories5

  • Archives67

  • resume & publications

  • links

【论文阅读】Spatio-Temporal Filter Adaptive Network for Video Deblurring

Posted on 2020-03-12 | In paper reading | Views:
Symbols count in article: 297 | Reading time ≈ 1 mins.

ICCV2019论文,旨在视频deblur。网站有论文讲解。论文使用STFAN用于视频去模糊,它集成帧对齐、去模糊到同一个框架中,而无需明显的运动估计,并且取得了SOTA结果。 pipeline 上图给出了文中所用到的网络架构示意图。从中可以看出,它包含三个子模块:特征提取、STFAN以及重建模块。 ...

Read more »

【论文阅读】Reblur2Deblur:Deblurring Videos via Self-Supervised Learning

Posted on 2020-03-12 | In paper reading | Views:
Symbols count in article: 130 | Reading time ≈ 1 mins.

ICCP2018论文(好像还不在CCF名单里面???),用于自监督视频去模糊。 思想很简单—估计出blur kernel之后(基于光流),在reblur,用input做监督。 下图是基于光流估计blur-kernel,具有很好的参考价值。 具体细节相见论文。

Read more »

【论文阅读】Down-Scaling with Learned Kernels in Multi-Scale Deep Neural Networks for Non-Uniform Single Image Deblurring

Posted on 2020-03-12 | In paper reading | Views:
Symbols count in article: 260 | Reading time ≈ 1 mins.

arxiv2019论文,方案应用于NTIRE19比赛。CSDN有简单的论文解读。作者认为现在的多尺度方法直接暴力地使用bicubic下采样进行尺度图像的生成;但是这样的方式在实际中不是最优的,并且会造成边缘的模糊,不利于deblur. 因此论文提出使用基于学习的降采样kernel降低分辨率,同时可以 ...

Read more »

KPN-基于核估计的low-level论文

Posted on 2020-03-11 | Edited on 2020-03-12 | In paper reading | Views:
Symbols count in article: 14 | Reading time ≈ 1 mins.

【coming soon】

Read more »

【论文阅读】Deep Blind Video Super-resolution

Posted on 2020-03-11 | In paper reading | Views:
Symbols count in article: 58 | Reading time ≈ 1 mins.

arxiv202003论文,可能中了CVPR2020。论文使用kernel-modeling的方法处理VSR问题。

Read more »

【论文阅读】Model-blind Video Denoising Via Frame-to-frame Training

Posted on 2020-03-11 | In paper reading | Views:
Symbols count in article: 15 | Reading time ≈ 1 mins.

CVPR2019论文。有讲解

Read more »

【论文阅读】Restore from Restored:Video Restoration with Pseudo Clean Video

Posted on 2020-03-11 | Edited on 2020-03-12 | In paper reading | Views:
Symbols count in article: 269 | Reading time ≈ 1 mins.

arxiv202003论文,投稿ECCV2020。比较新的一篇自监督视频low-level(denoise)的论文。 related work论文指出,用于图像恢复的常见自然图像属性是patch-recurrence,也就是在不同scale中存在相似的patches,这样的方法在SISR被大量采用[ ...

Read more »

【论文阅读】Bringing Alive Blurred Moments

Posted on 2020-03-09 | Edited on 2020-03-12 | In paper reading | Views:
Symbols count in article: 27 | Reading time ≈ 1 mins.

ICCV2019论文。【coming soons】

Read more »

GCN

Posted on 2020-03-04 | Edited on 2020-03-09 | In theoretical analysis | Views:
Symbols count in article: 12 | Reading time ≈ 1 mins.

coming soon

Read more »

【论文阅读】DAVID:Dual-Attentional Video Deblurring

Posted on 2020-03-01 | In paper reading | Views:
Symbols count in article: 134 | Reading time ≈ 1 mins.

WACV2020的论文。论文认为,相机抖动通常会导致短期、空间均匀且在时间上不相关的模糊,而物体运动会导致长期,空间局部且在时间上平滑的模糊。这种时空层面的“异质性”并不能很好的表示(现有方法都是输入固定的几帧视频)。因此论文设计内部和外部的attention模块。

Read more »
123…7

Zeyuxiao

67 posts
5 categories
48 tags
RSS
GitHub E-Mail
Links
  • USTC official website
  • HFUT official website
© 2019 – 2020 Zeyuxiao | 40k | 50 mins.
Powered by Hexo v3.8.0
|
Theme – NexT.Mist v7.1.0
|