Action recognition

MAR: Masked Autoencoders for Efficient Action Recognition

[TMM] Efficient masked action recognition model that reduces the computation of ViT by 53% with better performances.

Zhiwu Qing, Shiwei Zhang, Ziyuan Huang, Xiang Wang, Yuehuan Wang, Yiliang Lv, Changxin Gao, Nong Sang

MAR: Masked Autoencoders for Efficient Action Recognition

Learning from Untrimmed Videos: Self-Supervised Video Representation Learning with Hierarchical Consistency

[CVPR 2022] A contrasive learning framework leveraging rich information in long videos.

Zhiwu Qing, Shiwei Zhang, Ziyuan Huang, Yi Xu, Xiang Wang, Mingqian Tang, Changxin Gao, Rong Jin, Nong Sang

Learning from Untrimmed Videos: Self-Supervised Video Representation Learning with Hierarchical Consistency

ParamCrop: Parametric Cubic Cropping for Video Contrastive Learning

[TMM] Parameterized cropping module for better contrastive learning in videos.

Zhiwu Qing, Ziyuan Huang, Shiwei Zhang, Mingqian Tang, Changxin Gao, Marcelo H. Ang Jr, Rong Jin, Nong Sang

ParamCrop: Parametric Cubic Cropping for Video Contrastive Learning