Learning Graph Convolutional Networks for Multi-Label Recognition and Applications

被引:55
作者
Chen, Zhao-Min [1 ,2 ]
Wei, Xiu-Shen [3 ,4 ]
Wang, Peng [5 ]
Guo, Yanwen [1 ]
机构
[1] Nanjing Univ, Natl Key Lab Novel Software Technol, Nanjing 210023, Peoples R China
[2] Megvii Technol, Megvii Res Nanjing, Nanjing 210000, Peoples R China
[3] Nanjing Univ Sci & Technol, Sch Comp Sci & Engn, PCA Lab,Minist Educ, Key Lab Intelligent Percept & Syst High Dimens Inf, Nanjing 210094, Peoples R China
[4] Nanjing Univ Sci & Technol, Sch Comp Sci & Engn, Jiangsu Key Lab Image & Video Understanding Social, Nanjing 210094, Peoples R China
[5] Univ Wollongong, Sydney, NSW 2170, Australia
基金
中国国家自然科学基金;
关键词
Image recognition; Correlation; Face recognition; Task analysis; Semantics; Topology; Computational modeling; Multi-label recognition; graph convolutional networks; convolutional neural networks; label dependency; IMAGE; CLASSIFICATION;
D O I
10.1109/TPAMI.2021.3063496
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The task of multi-label image recognition is to predict a set of object labels that present in an image. As objects normally co-occur in an image, it is desirable to model the label dependencies to improve the recognition performance. To capture and explore such important information, we propose graph convolutional networks (GCNs) based models for multi-label image recognition, where directed graphs are constructed over classes and information is propagated between classes to learn inter-dependent class-level representations. Following this idea, we design two particular models that approach multi-label classification from different views. In our first model, the prior knowledge about the class dependencies is integrated into classifier learning. Specifically, we propose Classifier Learning GCN (C-GCN) to map class-level semantic representations (e.g., word embeddings) into classifiers that maintain the inter-class topology. In our second model, we decompose the visual representation of an image into a set of label-aware features and propose prediction learning GCN (P-GCN) to encode such features into inter-dependent image-level prediction scores. Furthermore, we also present an effective correlation matrix construction approach to capture inter-class relationships and consequently guide information propagation among classes. Empirical results on generic multi-label image recognition demonstrate that both of the proposed models can obviously outperform other existing state-of-the-arts. Moreover, the proposed methods also show advantages in some other multi-label classification related applications.
引用
收藏
页码:6969 / 6983
页数:15
相关论文
共 64 条
[1]   Label-Embedding for Image Classification [J].
Akata, Zeynep ;
Perronnin, Florent ;
Harchaoui, Zaid ;
Schmid, Cordelia .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2016, 38 (07) :1425-1438
[2]  
Bhattarai B, 2019, Arxiv, DOI arXiv:1907.06757
[3]  
Bruna J, 2014, Arxiv, DOI arXiv:1312.6203
[4]  
Chen TS, 2018, AAAI CONF ARTIF INTE, P6730
[5]   Multi-Label Image Recognition with Graph Convolutional Networks [J].
Chen, Zhao-Min ;
Wei, Xiu-Shen ;
Wang, Peng ;
Guo, Yanwen .
2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, :5172-5181
[6]   Xception: Deep Learning with Depthwise Separable Convolutions [J].
Chollet, Francois .
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017), 2017, :1800-1807
[7]  
Chua T-S, 2009, P ACM INT C IM VID R, P1, DOI DOI 10.1145/1646396.1646452
[8]  
Defferrard M, 2016, ADV NEUR IN, V29
[9]   Scalable Multi-label Annotation [J].
Deng, Jia ;
Russakovsky, Olga ;
Krause, Jonathan ;
Bernstein, Michael S. ;
Berg, Alex ;
Li Fei-Fei .
32ND ANNUAL ACM CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS (CHI 2014), 2014, :3099-3102
[10]  
Deng J, 2009, PROC CVPR IEEE, P248, DOI 10.1109/CVPRW.2009.5206848