Hierarchical compositional representations for few-shot action recognition

doi:10.1016/j.cviu.2023.103911

	Hierarchical compositional representations for few-shot action recognition
	Li, Changzhen 1,2,3; Zhang, Jie 1,2; Wu, Shuzhe 4; Jin, Xin 4; Shan, Shiguang 1,2,3
	2024-03-01
发表期刊	COMPUTER VISION AND IMAGE UNDERSTANDING
ISSN	1077-3142
卷号	240 页码:11
摘要	Recently action recognition has received more and more attention for its comprehensive and practical applications in intelligent surveillance and human-computer interaction. However, few-shot action recognition has not been well explored and remains challenging because of data scarcity. In this paper, we propose a novel hierarchical compositional representations (HCR) learning approach for few-shot action recognition. Specifically, we divide a complicated action into several sub-actions by carefully designed hierarchical clustering and further decompose the sub-actions into more fine-grained spatially attentional sub-actions (SASactions). Although there exist large differences between base classes and novel classes, they can share similar patterns in sub-actions or SAS-actions. Furthermore, we adopt the Earth Mover's Distance in the transportation problem to measure the similarity between video samples in terms of sub-action representations. It computes the optimal matching flows between sub-actions as distance metric, which is favorable for comparing finegrained patterns. Extensive experiments show our method achieves the state-of-the-art results on HMDB51, UCF101 and Kinetics datasets.
关键词	Action recognition Few-shot learning Hierarchical compositional representations Body parts EMD distance
DOI	10.1016/j.cviu.2023.103911
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China[2021YFC3310100] ; National Natural Science Foundation of China[62176251] ; Youth Innovation Promotion Association CAS
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:001166488900001
出版者	ACADEMIC PRESS INC ELSEVIER SCIENCE
引用统计	被引频次：3[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/38843
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Zhang, Jie
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Key Lab Intelligent Informat Proc, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China 3.UCAS, Hangzhou Inst Adv Study, Sch Intelligent Sci & Technol, Hangzhou, Peoples R China 4.Beijing Huawei Cloud Comp Technol Co Ltd, 3 Xinxi Rd, Beijing 100095, Peoples R China
推荐引用方式 GB/T 7714	Li, Changzhen,Zhang, Jie,Wu, Shuzhe,et al. Hierarchical compositional representations for few-shot action recognition[J]. COMPUTER VISION AND IMAGE UNDERSTANDING,2024,240:11.
APA	Li, Changzhen,Zhang, Jie,Wu, Shuzhe,Jin, Xin,&Shan, Shiguang.(2024).Hierarchical compositional representations for few-shot action recognition.COMPUTER VISION AND IMAGE UNDERSTANDING,240,11.
MLA	Li, Changzhen,et al."Hierarchical compositional representations for few-shot action recognition".COMPUTER VISION AND IMAGE UNDERSTANDING 240(2024):11.