论文题目:Hierarchically Self-Supervised Transformer for Human Skeleton Representation Learning
论文下载地址:https://www.ecva.net/papers/eccv_2022/papers_ECCV/papers/136860181.pdf
代码地址:https://github.com/yuxiaochen1103…
目录
Contributions
Method
Cross-Modal Loss
Diversity Loss
Combining Both Loss Contributions
Results 论文名称:Cross and Learn: Cross-Modal Self-supervision(2018 GCPR: German Conference on Pattern Recognition)
论文作者…
On Measuring and Controlling the Spectral Bias of the Deep Image Prior 文章目录 On Measuring and Controlling the Spectral Bias of the Deep Image Prior1. 方法原理1.1 动机1.2 相关概念1.3 方法原理频带一致度量与网络退化谱偏移和网络结构的关系Lipschitz-controlle…
写在前面
方法部分好多公式变量,编辑器再打一遍好麻烦。。。偷个懒,就直接把这部分的笔记导出成图片咯,并且我按我自己理解比较顺的逻辑重新捋了一下。 3D Human Action Representation Learning via Cross-View Consistency Pursuit
&…
论文名称:Self-supervised video representation learning with space-time cubic puzzles(2019 AAAI)
论文作者:Dahun Kim, Donghyeon Cho, In So Kweon
下载地址:https://ojs.aaai.org/index.php/AAAI/article/vie…
目录
Contributions
Method
1、Objective Function
2、Class rebalancing
3、Class Probabilities to Point Estimates
Code
Results
1、Colorization results on 10k images in the ImageNet validation set
2、Colorization Turing Test.
3、Cross-Channel Encodin…
Efficient Self-supervised Learning with Contextualized Target Representations for Vision, Speech and Language
(视觉、语音和语言的语境化目标表征的高效自监督学习) 论文:efficient-self-supervised-learning-with-contextualized-t…
论文名称:Emerging Properties in Self-Supervised Vision Transformers 发表时间:CVPR2021 作者及组织: Facebook AI Research GitHub:https://github.com/facebookresearch/dino/tree/main
问题与贡献
作者认为self-supervise…