文献类型：期刊文献

英文题名：ArCycleGAN: Improved CycleGAN for Style Transferring of Fruit Images

作者：Chen, Hongqian[1,3];Guan, Mengxi[1,3];Li, Hui[2]

第一作者：Chen, Hongqian

通讯作者：Li, H[1]

机构：[1]Beijing Technol & Business Univ, Sch Comp Sci & Engn, Beijing 100048, Peoples R China;[2]Beijing Union Univ, Management Coll, Beijing 100101, Peoples R China;[3]Beijing Technol & Business Univ, Beijing Key Lab Big Data Technol Food Safety, Beijing 100048, Peoples R China

第一机构：Beijing Technol & Business Univ, Sch Comp Sci & Engn, Beijing 100048, Peoples R China

通讯机构：[1]corresponding author), Beijing Union Univ, Management Coll, Beijing 100101, Peoples R China.|[1141755]北京联合大学管理学院;[11417]北京联合大学;

年份：2021

卷号：9

起止页码：46776-46787

外文期刊名：IEEE ACCESS

收录：;EI(收录号：20211310151117);Scopus(收录号：2-s2.0-85103288955);WOS:【SCI-EXPANDED(收录号:WOS:000637167800001)】；

基金：This work was supported in part by the National Natural Science Foundation of China under Grant 31701517, in part by the Beijing Philosophy and Social Science Foundation under Grant 17GLC060 and Grant 20GLB032, and in part by the Academic Research Projects of Beijing Union University under Grant ZB10202005.

语种：英文

外文关键词：Training; Convolution; Generators; Image reconstruction; Image recognition; Generative adversarial networks; Diseases; Image generation; style transferring; attribute registration; image registration; improved CycleGAN

摘要：CycleGAN can realize image translation and style transferring among unpaired images. However, it will easily generate inappropriate image results when the number and shapes of objects in the style offering image and the source image are greatly different. The paper proposed an improved network, named arCycleGAN, which introduced the mechanism of attribute registration into CycleGAN to solve the problem. The arCycleGAN can transfer the freshness styles from the style offering images to the unpaired input source images. The generated target images will have the freshness attributes of the style offering images, while maintaining the shapes and key features of the input source images. The realization of mechanism of attribute registration consists of three modules. The first module is attribute recognition module, which can identify and label the attributes of objects in images. The second module is image pre-screening module, which selects appropriate image subset as screened training set from raw image set according to the attributes of the input source images. The third module is similarity matching module, which matches the images in screened training set based on the similarity. The generator and discriminator in the new network are similar to that in the CycleGAN network. Experimental results demonstrate the effectiveness and better performance of the arCycleGAN. Compared with the CycleGAN, the new network can generate more convincing images. It can generate the target images of similar quality based on a smaller training set and less training time than the original CycleGAN. For generating images of similar quality, the number of images in the required training set can be reduced by 50%, while training time is reduced by 5.8%.