Open AccessAnalysis of Image Generation by different Generator in GANs
Author(s) -
Jiabei He,
Yuzheng Nie,
Mao Zi-wen
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1903/1/012061
Subject(s) - generator (circuit theory) , computer science , image (mathematics) , artificial intelligence , field (mathematics) , third generation , image quality , computer vision , pattern recognition (psychology) , mathematics , telecommunications , power (physics) , physics , quantum mechanics , pure mathematics
GAN is very useful in the field of image generation. Many related GANs have been proposed for generating images from the description. However, the research about analysis of image generation by different Generator in GANs is still insufficient. In this paper, different methods such as CNN and Resnet are used as the Generator in GANs for image generation. The subjective evaluation is used to analyze the performance of different generators of GAN. The result shows that the CNN-based generator performs better than the Resnet-based generator. And the increase of the number of parameters in the model will change the image quality. The analysis of different methods as the generator in GAN has a great referential significance in the field of image generation.