GCAM: lightweight image inpainting via group convolution and attention mechanism

被引:21
作者
Chen, Yuantao [1 ,2 ]
Xia, Runlong [3 ]
Yang, Kai [4 ]
Zou, Ke [5 ]
机构
[1] Hunan Univ Informat Technol, Sch Comp Sci & Engn, Changsha 410151, Hunan, Peoples R China
[2] Hunan Univ Informat Technol, Hunan Prov Higher Educ Key Lab Intelligent Sensing, Changsha 410151, Hunan, Peoples R China
[3] Mt Yuelu Breeding Innovat Ctr Ltd, Changsha 410000, Peoples R China
[4] Hunan ZOOMLION Intelligent Technol Corp Ltd, Changsha 410005, Hunan, Peoples R China
[5] Hunan Initial New Mat Corp Ltd, Loudi 417000, Peoples R China
关键词
Image inpainting; Deep learning; Lightweight network; Group convolution; Attention mechanism;
D O I
10.1007/s13042-023-01999-z
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recently, image inpainting techniques tend to be more concerned with how to enhance the quality of restoration than with how to function on various platforms with limited processing power. In this paper, we propose a lightweight method that combines group convolution and attention mechanism to improve or replace the traditional convolution module. Group convolution was used to achieve multi-level image inpainting, and the authors proposed the rotating attention mechanism for allocation to deal with the issue of information mobility between channels in traditional convolution processing. The parallel discriminator structure was utilized throughout the network's overall design phase to guarantee both local and global consistency of the image inpainting process. The experimental results can demonstrate that, while the quality of image inpainting has been ensured, the proposed image inpainting network's inference time and resource usage are significantly lower than those of comparable lightweight approaches.
引用
收藏
页码:1815 / 1825
页数:11
相关论文
共 31 条
[1]   Distracted driver detection using compressed energy efficient convolutional neural network [J].
Alzubi, Jafar A. ;
Jain, Rachna ;
Alzubi, Omar ;
Thareja, Anuj ;
Upadhyay, Yash .
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS, 2022, 42 (02) :1253-1265
[2]  
Alzubi JA, 2016, INT J ADV COMPUT SC, V7, P524
[3]   An optimal pruning algorithm of classifier ensembles: dynamic programming approach [J].
Alzubi, Omar A. ;
Alzubi, Jafar A. ;
Alweshah, Mohammed ;
Qiqieh, Issa ;
Al-Shami, Sara ;
Ramachandran, Manikandan .
NEURAL COMPUTING & APPLICATIONS, 2020, 32 (20) :16091-16107
[4]  
Broad T, 2016, SIGGRAPH 16, p54: 1
[5]   Mathematical models for local nontexture inpaintings [J].
Chan, TF ;
Shen, JH .
SIAM JOURNAL ON APPLIED MATHEMATICS, 2002, 62 (03) :1019-1043
[6]  
Chen Q., 2019, ARXIV
[7]   Region filling and object removal by exemplar-based image inpainting [J].
Criminisi, A ;
Pérez, P ;
Toyama, K .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2004, 13 (09) :1200-1212
[8]   What Makes Paris Look like Paris? [J].
Doersch, Carl ;
Singh, Saurabh ;
Gupta, Abhinav ;
Sivic, Josef ;
Efros, Alexei A. .
ACM TRANSACTIONS ON GRAPHICS, 2012, 31 (04)
[9]  
Grossauer H, 2004, LECT NOTES COMPUT SC, V3022, P214
[10]  
Hays J, 2007, ACM T GRAPHIC, V26, DOI [10.1145/1239451.1239455, 10.1145/1276377.1276382]