Abstract
The generative adversarial network (GAN) is first proposed in 2014, and this kind of network model is machine learning systems that can learn to measure a given distribution of data, one of the most important applications is style transfer. Style transfer is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image. CYCLE-GAN is a classic GAN model, which has a wide range of scenarios in style transfer. Considering its unsupervised learning characteristics, the mapping is easy to be learned between an input image and an output image. However, it is difficult for CYCLE-GAN to converge and generate high-quality images. In order to solve this problem, spectral normalization is introduced into each convolutional kernel of the discriminator. Every convolutional kernel reaches Lipschitz stability constraint with adding spectral normalization and the value of the convolutional kernel is limited to [0, 1], which promotes the training process of the proposed model. Besides, we use pretrained model (VGG16) to control the loss of image content in the position of l1 regularization. To avoid overfitting, l1 regularization term and l2 regularization term are both used in the object loss function. In terms of Frechet Inception Distance (FID) score evaluation, our proposed model achieves outstanding performance and preserves more discriminative features. Experimental results show that the proposed model converges faster and achieves better FID scores than the state of the art.
Highlights
Goodfellow et al [1,2] proposed a new neural network model in 2014, and named it as generative adversarial network (GAN)
The designed generator firstly extracts the image features through a three-layer convolutional neural network, learns the image features through a nine-layer residual network, and reconstructs the image features through a three-layer deconvolutional network
On the basis of CYCLE-GAN, the content consistency loss function is proposed with the help of pretrained model (VGG16)
Summary
Goodfellow et al [1,2] proposed a new neural network model in 2014, and named it as generative adversarial network (GAN). One of the most important is to solve the problem of high resolution [27] and multi-target style transfer To achieve this goal, some researchers propose new methods. The main methods are based on the GAN On this basis, adjusting the architecture [28] and reconstructing the object loss function is necessary. Our proposed model achieves style transfer based on unpaired images and uses lightweight neural networks to generate better results. The discriminator relies on the method of embedded normalization [31,32,33,34,35,36,37,38], and it reduces the oscillation of the object loss function during model convergence.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
