中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 自动化研究所
  3. 中国科学院自动化研究所
  4. 多模态人工智能系统全国重点实验室
Attention Weighted Local Descriptors

文献类型：期刊论文

作者Wang, Changwei1,6; Xu, Rongtao1,6; Lu, Ke5; Xu, Shibiao4; Meng, Weiliang1,6; Zhang, Yuyang2; Fan, Bin3; Zhang, Xiaopeng1,6
刊名IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
出版日期2023-09-01
卷号45期号:9页码:10632-10649
ISSN号0162-8828
关键词Local features detection and description consistent attention mechanism context augmentation lightweight local descriptors knowledge distillation
DOI10.1109/TPAMI.2023.3266728
通讯作者Xu, Shibiao(shibiaoxu@bupt.edu.cn) ; Zhang, Xiaopeng(Xiaopeng.Zhang@ia.ac.cn)
英文摘要Local features detection and description are widely used in many vision applications with high industrial and commercial demands. With large-scale applications, these tasks raise high expectations for both the accuracy and speed of local features. Most existing studies on local features learning focus on the local descriptions of individual keypoints, which neglect their relationships established from global spatial awareness. In this paper, we present AWDesc with a consistent attention mechanism (CoAM) that opens up the possibility for local descriptors to embrace image-level spatial awareness in both the training and matching stages. For local features detection, we adopt local features detection with feature pyramid to obtain more stable and accurate keypoints localization. For local features description, we provide two versions of AWDesc to cope with different accuracy and speed requirements. On the one hand, we introduce Context Augmentation to address the inherent locality of convolutional neural networks by injecting non-local context information, so that local descriptors can "look wider to describe better". Specifically, well-designed Adaptive Global Context Augmented Module (AGCA) and Diverse Surrounding Context Augmented Module (DSCA) are proposed to construct robust local descriptors with context information from global to surrounding. On the other hand, we design an extremely lightweight backbone network coupled with the proposed special knowledge distillation strategy to achieve the best trade-off in accuracy and speed. What is more, we perform thorough experiments on image matching, homography estimation, visual localization, and 3D reconstruction tasks, and the results demonstrate that our method surpasses the current state-of-the-art local descriptors. Code is available at: https://github.com/vignywang/AWDesc.
资助项目National Natural Science Foundation of China[U21A20515] ; National Natural Science Foundation of China[62032022] ; National Natural Science Foundation of China[62271074] ; National Natural Science Foundation of China[62222302] ; National Natural Science Foundation of China[U2003109] ; National Natural Science Foundation of China[62171321] ; National Natural Science Foundation of China[62071157] ; National Natural Science Foundation of China[62162044] ; Open Research Fund of Key Laboratory of Space Utilization, Chinese Academy of Sciences[LSU-KFJJ-2021-05] ; Open Projects Program of National Laboratory of Pattern Recognition
WOS研究方向Computer Science ; Engineering
语种英语
出版者IEEE COMPUTER SOC
WOS记录号WOS:001045832200003
资助机构National Natural Science Foundation of China ; Open Research Fund of Key Laboratory of Space Utilization, Chinese Academy of Sciences ; Open Projects Program of National Laboratory of Pattern Recognition
源URL[http://ir.ia.ac.cn/handle/173211/53894]  
专题多模态人工智能系统全国重点实验室
通讯作者Xu, Shibiao; Zhang, Xiaopeng
作者单位1.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100045, Peoples R China
2.Thunder Software Technol Co Ltd, Chengdu 610000, Peoples R China
3.Univ Sci & Technol Beijing, Sch Intelligence Sci & Technol, Beijing 100083, Peoples R China
4.Beijing Univ Posts & Telecommun, Sch Artificial Intelligence, Beijing 100876, Peoples R China
5.Univ Chinese Acad Sci, Sch Engn Sci, Beijing 100049, Peoples R China
6.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100190, Peoples R China
推荐引用方式
GB/T 7714		 Wang, Changwei,Xu, Rongtao,Lu, Ke,et al. Attention Weighted Local Descriptors[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,2023,45(9):10632-10649.
APA Wang, Changwei.,Xu, Rongtao.,Lu, Ke.,Xu, Shibiao.,Meng, Weiliang.,...&Zhang, Xiaopeng.(2023).Attention Weighted Local Descriptors.IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,45(9),10632-10649.
MLA Wang, Changwei,et al."Attention Weighted Local Descriptors".IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 45.9(2023):10632-10649.

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。