{"title":"基于生成对抗网络的中国传统窗纱图像生成","authors":"Chengxi Miao, Jianqin Wu, Jialin Chen, Shiyi Xiong, Lingyue Wang, Qi Wang","doi":"10.1109/CoST57098.2022.00055","DOIUrl":null,"url":null,"abstract":"Window grille is one of the expressions of traditional Chinese folk arts, which has unique stylistic characteristics and rich symbolic meanings. Studying how to extract the style characteristics of window grilles and generate new window grilles is beneficial to the inheritance and development of traditional Chinese arts. In recent years, the innovative development of generative adversarial networks (GANs) has made it possible to capture the intrinsic distribution of data and generate images that look like real ones. On the basis of researching existing three types of style-based generative adversarial networks (StyleGANs) and adaptive discriminator augmentation (ADA), we use StyleGAN2-ADA to train window grille datasets and generate new window grille images. Finally, multiple image quality evaluation metrics are used to analyze the generated images. The result shows that StyleGAN2-ADA has a good effect on the automatic generation of window grille images. In addition, by comparing the results of different size datasets, we found that the size of dataset has a significant impact on the quality of the generated window grilles.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Image generation of traditional Chinese window grilles based on generative adversarial networks\",\"authors\":\"Chengxi Miao, Jianqin Wu, Jialin Chen, Shiyi Xiong, Lingyue Wang, Qi Wang\",\"doi\":\"10.1109/CoST57098.2022.00055\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Window grille is one of the expressions of traditional Chinese folk arts, which has unique stylistic characteristics and rich symbolic meanings. Studying how to extract the style characteristics of window grilles and generate new window grilles is beneficial to the inheritance and development of traditional Chinese arts. In recent years, the innovative development of generative adversarial networks (GANs) has made it possible to capture the intrinsic distribution of data and generate images that look like real ones. On the basis of researching existing three types of style-based generative adversarial networks (StyleGANs) and adaptive discriminator augmentation (ADA), we use StyleGAN2-ADA to train window grille datasets and generate new window grille images. Finally, multiple image quality evaluation metrics are used to analyze the generated images. The result shows that StyleGAN2-ADA has a good effect on the automatic generation of window grille images. In addition, by comparing the results of different size datasets, we found that the size of dataset has a significant impact on the quality of the generated window grilles.\",\"PeriodicalId\":135595,\"journal\":{\"name\":\"2022 International Conference on Culture-Oriented Science and Technology (CoST)\",\"volume\":\"25 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-08-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2022 International Conference on Culture-Oriented Science and Technology (CoST)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/CoST57098.2022.00055\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CoST57098.2022.00055","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Image generation of traditional Chinese window grilles based on generative adversarial networks
Window grille is one of the expressions of traditional Chinese folk arts, which has unique stylistic characteristics and rich symbolic meanings. Studying how to extract the style characteristics of window grilles and generate new window grilles is beneficial to the inheritance and development of traditional Chinese arts. In recent years, the innovative development of generative adversarial networks (GANs) has made it possible to capture the intrinsic distribution of data and generate images that look like real ones. On the basis of researching existing three types of style-based generative adversarial networks (StyleGANs) and adaptive discriminator augmentation (ADA), we use StyleGAN2-ADA to train window grille datasets and generate new window grille images. Finally, multiple image quality evaluation metrics are used to analyze the generated images. The result shows that StyleGAN2-ADA has a good effect on the automatic generation of window grille images. In addition, by comparing the results of different size datasets, we found that the size of dataset has a significant impact on the quality of the generated window grilles.