LARNeXt: End-to-End Lie Algebra Residual Network for Face Recognition
文献类型:期刊论文
作者 | Yang, Xiaolong2; Jia, Xiaohong2; Gong, Dihong4; Yan, Dong-Ming1,3,5; Li, Zhifeng4; Liu, Wei4 |
刊名 | IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE |
出版日期 | 2023-10-01 |
卷号 | 45期号:10页码:11961-11976 |
ISSN号 | 0162-8828 |
关键词 | Face recognition lie algebra pose estimation profile face |
DOI | 10.1109/TPAMI.2023.3279378 |
通讯作者 | Jia, Xiaohong(xhjia@amss.ac.cn) ; Li, Zhifeng(michaelzfli@tencent.com) ; Liu, Wei(wl2223@columbia.edu) |
英文摘要 | Face recognition has always been courted in computer vision and is especially amenable to situations with significant variations between frontal and profile faces. Traditional techniques make great strides either by synthesizing frontal faces from sizable datasets or by empirical pose invariant learning. In this paper, we propose a completely integrated embedded end-to-end Lie algebra residual architecture (LARNeXt) to achieve pose robust face recognition. First, we explore how the face rotation in the 3D space affects the deep feature generation process of convolutional neural networks (CNNs), and prove that face rotation in the image space is equivalent to an additive residual component in the feature space of CNNs, which is determined solely by the rotation. Second, on the basis of this theoretical finding, we further design three critical subnets to leverage a soft regression subnet with novel multi-fusion attention feature aggregation for efficient pose estimation, a residual subnet for decoding rotation information from input face images, and a gating subnet to learn rotation magnitude for controlling the strength of the residual component that contributes to the feature learning process. Finally, we conduct a large number of ablation experiments, and our quantitative and visualization results both corroborate the credibility of our theory and corresponding network designs. Our comprehensive experimental evaluations on frontal-profile face datasets, general unconstrained face recognition datasets, and industrial-grade tasks demonstrate that our method consistently outperforms the state-of-the-art ones. |
资助项目 | National Key Research and Development Program ; CAS Project[2019YFB2204104] ; National Natural Science Foundation of China[YSBR-034] ; National Natural Science Foundation of China[12022117] ; Tencent AI Lab Rhino-Bird Focused Research Program[62172415] ; [JR202127] |
WOS研究方向 | Computer Science ; Engineering |
语种 | 英语 |
出版者 | IEEE COMPUTER SOC |
WOS记录号 | WOS:001068816800029 |
资助机构 | National Key Research and Development Program ; CAS Project ; National Natural Science Foundation of China ; Tencent AI Lab Rhino-Bird Focused Research Program |
源URL | [http://ir.ia.ac.cn/handle/173211/53050] |
专题 | 多模态人工智能系统全国重点实验室 |
通讯作者 | Jia, Xiaohong; Li, Zhifeng; Liu, Wei |
作者单位 | 1.Chinese Acad Sci, Inst Automat, Key Lab Multimodal Artificial Intelligence Syst MA, Beijing 101408, Peoples R China 2.Univ Chinese Acad Sci, Chinese Acad Sci, Acad Math & Syst Sci, Huairou 101408, Peoples R China 3.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 101408, Peoples R China 4.Tencent Data Platform, Shenzhen 518054, Peoples R China 5.Chinese Acad Sci, Inst Automat, NLPR, Beijing 101408, Peoples R China |
推荐引用方式 GB/T 7714 | Yang, Xiaolong,Jia, Xiaohong,Gong, Dihong,et al. LARNeXt: End-to-End Lie Algebra Residual Network for Face Recognition[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,2023,45(10):11961-11976. |
APA | Yang, Xiaolong,Jia, Xiaohong,Gong, Dihong,Yan, Dong-Ming,Li, Zhifeng,&Liu, Wei.(2023).LARNeXt: End-to-End Lie Algebra Residual Network for Face Recognition.IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,45(10),11961-11976. |
MLA | Yang, Xiaolong,et al."LARNeXt: End-to-End Lie Algebra Residual Network for Face Recognition".IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 45.10(2023):11961-11976. |
入库方式: OAI收割
来源:自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。