Spectral Normalized CycleGAN with Application in Semisupervised Semantic Segmentation of Sonar Images. (28th April 2022)

• Record Type: Journal Article
• Authors: Zhang, Zhisheng; Tang, Jinsong; Zhong, Heping; Wu, Haoran; Zhang, Peng; Ning, Mingqiang
• Other Names: Loddo Andrea Academic Editor.
• Is Part Of: Computational intelligence and neuroscience. Volume 2022(2022)
• Publication Date: 2022-04-28
• Subjects: Neurosciences -- Data processing -- Periodicals; Computational intelligence -- Periodicals; Computational neuroscience -- Periodicals; 612.80285
• DOI: 10.1155/2022/1274260
• Languages: English
• ISSNs: 1687-5265

Abstract

Abstract : The effectiveness of CycleGAN is demonstrated to outperform recent approaches for semisupervised semantic segmentation on public segmentation benchmarks. In contrast to analog images, however, the acoustic images are unbalanced and often exhibit speckle noise. As a consequence, CycleGAN is prone to mode-collapse and cannot retain target details when applied directly to the sonar image dataset. To address this problem, a spectral normalized CycleGAN network is presented, which applies spectral normalization to both generators and discriminators to stabilize the training of GANs. Without using a pretrained model, the experimental results demonstrate that our simple yet effective method helps to achieve reasonably accurate sonar targets segmentation results.