在这个步骤中,骨架增强器 R 会对输入的骨架数据 X进行增强。具体来说,MLP(多层感知器)生成一个四元数(Quaternion),这个四元数用于旋转骨架数据,改变它的视点,从而生成增强后的骨架数据。增强后的骨架数据同样通过GCN模型 f 提取特征。此时,骨架增强器通过最小化原始特征和增强特征之间的互信息来生成更具挑战性的...
Self-supervised action recognition plays a crucial role by enabling machines to understand and interpret human actions without the need for numerous human-annotated labels. Contrastive learning, which compels the model to focus on discriminative features by constructing positive and negative sample pairs,...
1DMMG: Dual Min-Max Games for Self-SupervisedSkeleton-Based Action RecognitionShannan GUAN ∗ , Xin YU † , Wei HUANG ‡ , Gengfa FANG § , and Haiyan LU ∗∗Australia Artif i cial Intelligence InstituteUniversity of Technology Sydney, Australia, AU† School of Information Technology...
Recently, self-supervised learning (SSL) has been proved very effective and it can help boost the performance in learning representations from unlabeled data in the image domain. Yet, very little is explored about its usefulness in 3D skeleton-based action recognition understanding. Directly applying...
* 题目: Contrastive Learning from Extremely Augmented Skeleton Sequences for Self-supervised Action Recognition* 链接: arxiv.org/abs/2112.0359* 作者: Tianyu Guo,Hong Liu,Zhan Chen,Mengyuan Liu,Tao Wang,Runwei Ding* 其他: Accepted by AAAI2022* 摘要: 近年来,随着对比学习方法的进步,基于骨架的动作...
Paper tables with annotated results for Cross-Stream Contrastive Learning for Self-Supervised Skeleton-Based Action Recognition
Contribute to sadimanna/self-supervised-learning-and-contrastive-learning-papers development by creating an account on GitHub.
《Channel-wise Topology Refinement Graph Convolution for Skeleton-Based Action Recognition》(ICCV 2021) GitHub:https:// github.com/Uason-Chen/CTR-GCN [fig6]《A Transformer-based Framework for Multivariate Time Series Representation Learning》(KDD 2021) GitHub:https:// github.com/gzerveas/mvts_...
In 3D action recognition, there exists rich complementary information between skeleton modalities. Nevertheless, how to model and utilize this information remains a challenging problem for self-supervised 3D action representation learning. In this work, we formulate the crossmodal interaction as a bidirect...
[ICCSNT 2021] Graph Data Augmentation based on Adaptive Graph Convolution for Skeleton-based Action Recognition [paper] [IJCNN 2021] Node Embedding using Mutual Information and Self-Supervision based Bi-level Aggregation [paper]Year 2020[Openreview 2020] Motif-Driven Contrastive Learning of Graph Represe...