IICNet: A Generic Framework for Reversible Image Conversion

被引:13
作者
Cheng, Ka Leong [1 ]
Xie, Yueqi [1 ]
Chen, Qifeng [1 ]
机构
[1] Hong Kong Univ Sci & Technol, Hong Kong, Peoples R China
来源
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) | 2021年
关键词
D O I
10.1109/ICCV48922.2021.00200
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Reversible image conversion (RIC) aims to build a reversible transformation between specific visual content (e.g., short videos) and an embedding image, where the original content can be restored from the embedding when necessary. This work develops Invertible Image Conversion Net (IICNet) as a generic solution to various RIC tasks due to its strong capacity and task-independent design. Unlike previous encoder-decoder based methods, IICNet maintains a highly invertible structure based on invertible neural networks (INNs) to better preserve the information during conversion. We use a relation module and a channel squeeze layer to improve the INN nonlinearity to extract cross-image relations and the network flexibility, respectively. Experimental results demonstrate that IICNet outperforms the specifically-designed methods on existing RIC tasks and can generalize well to various newly-explored tasks. With our generic IICNet, we no longer need to hand-engineer task-specific embedding networks for rapidly occurring visual content.
引用
收藏
页码:1971 / 1980
页数:10
相关论文
共 37 条
[1]   NTIRE 2017 Challenge on Single Image Super-Resolution: Dataset and Study [J].
Agustsson, Eirikur ;
Timofte, Radu .
2017 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION WORKSHOPS (CVPRW), 2017, :1122-1131
[2]  
[Anonymous], 2016, P INT C LEARN REPR I
[3]  
Ardizzone L., 2018, ARXIV180804730
[4]  
Ball Johannes, 2016, 5 INT C LEARNING REP
[5]  
Baluja Shumeet, 2017, Neural Information Processing Systems
[6]  
Bengio Y, 2015, 3 INT C LEARN REPR I
[7]  
Bengio Y., 2013, arXiv
[9]  
Dinh L., 2017, P 5 INT C LEARN REPR
[10]   ALIG Net: Partial-Shape Agnostic Alignment via Unsupervised Learning [J].
Hanocka, Rana ;
Fish, Noa ;
Wang, Zhenhua ;
Giryes, Raja ;
Fleishman, Shachar ;
Cohen-Or, Daniel .
ACM TRANSACTIONS ON GRAPHICS, 2019, 38 (01)