Enriched CNN-Transformer Feature Aggregation Networks for Super-Resolution

被引:30
作者
Yoo, Jinsu [1 ]
Kim, Taehoon [2 ]
Lee, Sihaeng [2 ]
Kim, Seung Hwan [2 ]
Lee, Honglak [2 ]
Kim, Tae Hyun [1 ]
机构
[1] Hanyang Univ, Seoul, South Korea
[2] LG AI Res, Seoul, South Korea
来源
2023 IEEE/CVF WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV) | 2023年
关键词
D O I
10.1109/WACV56688.2023.00493
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recent transformer-based super-resolution (SR) methods have achieved promising results against conventional CNN-based methods. However, these approaches suffer from essential shortsightedness created by only utilizing the standard self-attention-based reasoning. In this paper, we introduce an effective hybrid SR network to aggregate enriched features, including local features from CNNs and long-range multi-scale dependencies captured by transformers. Specifically, our network comprises transformer and convolutional branches, which synergetically complement each representation during the restoration procedure. Furthermore, we propose a cross-scale token attention module, allowing the transformer branch to exploit the informative relationships among tokens across different scales efficiently. Our proposed method achieves state-of-the-art SR results on numerous benchmark datasets.
引用
收藏
页码:4945 / 4954
页数:10
相关论文
共 51 条
[1]  
[Anonymous], 2018, Advances in Neural Information Processing Systems
[2]  
Ba J.L., 2016, ARXIV160706450, P1607
[3]   Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding [J].
Bevilacqua, Marco ;
Roumy, Aline ;
Guillemot, Christine ;
Morel, Marie-Line Alberi .
PROCEEDINGS OF THE BRITISH MACHINE VISION CONFERENCE 2012, 2012,
[4]  
Cao Jiezhang, 2021, ARXIV210606847
[5]  
Carion Nicolas, 2020, Computer Vision - ECCV 2020. 16th European Conference. Proceedings. Lecture Notes in Computer Science (LNCS 12346), P213, DOI 10.1007/978-3-030-58452-8_13
[6]  
Chen Chieh-Yun, 2021, P IEEE CVF INT C COM
[7]  
Chen Hanting, 2021, P IEEE CVF C COMP VI
[8]  
Chen Yinpeng, 2021, ARXIV210805895
[9]  
d'Ascoli Stephane, 2021, INT C MACH LEARN ICM
[10]   Second-order Attention Network for Single Image Super-Resolution [J].
Dai, Tao ;
Cai, Jianrui ;
Zhang, Yongbing ;
Xia, Shu-Tao ;
Zhang, Lei .
2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, :11057-11066