Various Generative Adversarial Networks Model for Synthetic Prohibitory Sign Image Generation
Round 1
Reviewer 1 Report
This paper reports a successful experiments of generating sign images of Taiwan. The contents are well described and easy to read.
However, the results are not surprising--which make this manuscript less interesting--as GAN is well known to the public and in this field. My comments are as follows:
- More analysis would be necessary on why LSGAN made better results.
- Network structure of the Generator/ Discriminator should be explained
- Results analysis according to the change of the network structure.
- Conclusion of the manuscript seems to be weak. It would be better to point out the conclusions clearly.
- What kind of database did the authors use to generate new images? If the pictures were taken by the authors, the procedures should be described as well.
Author Response
Dear Reviewer,
Many thanks for allowing us to revise our manuscript for possible publication in the Journal Applied Science. The paper is titled “Various Generative Adversarial Networks model for Synthetic Prohibitory Sign Image Generation." We have modified the manuscript according to your comments and the detailed corrections are listed in the attached file.
Regards,
Rung-Ching Chen
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper seems appropriate to measure traffic sign images generated by DCGAN, LSGAN, and WGAN with SSIM and MSE.
However, the paper lacks novelty generating traffic sign images only using existing GANs(DCGAN, LSGAN, and WGAN).
In addition to applying the existing GANs, designing a new optimized GAN to generate traffic sign images and comparing it with the existing GANs will be needed.
Author Response
Dear Reviewer,
Many thanks for allowing us to revise our manuscript for possible publication in the Journal Applied Science. The paper is titled “Various Generative Adversarial Networks model for Synthetic Prohibitory Sign Image Generation." We have modified the manuscript according to your comments and the detailed corrections are listed in the attached file.
Regards,
Rung-Ching Chen
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The topic of the manuscript is not much interesting yet, but manuscript has been well improved.
Reviewer 2 Report
Still, the novelty of the paper is judged to be insufficient.
However, it is judged that the overall method of experimenting using the existing GANs has performed well.
