Image-to-Image Local Feature Translation Using Double Adversarial Networks Based on CycleGAN

  • Chen WuEmail author
  • Lei Li
  • Zhenzhen Yang
  • Peihong Yan
  • Jiali Jiao
Conference paper
Part of the Lecture Notes in Electrical Engineering book series (LNEE, volume 516)


Image-to-image translation is a hot field in the machine learning with the emergency of the generative adversarial networks. Most of the latest models easily lead to changes in the overall image and overfitting when they are used to local feature translation. To address these limitations, this article adds a suppressor and proposes a double adversarial CycleGAN. The suppressor is added to suppress the change of images, and the suppressor and generator form a new adversarial relationship. We hope it will achieve Nash equilibrium that is the change of image focus on the local feature. Finally, a contrast experiment was conducted. In the case of image local feature transfer, the change of image is focused on the local features and the overfitting phenomenon can be well resolved.


Local feature translation Generative adversarial networks Double adversarial CycleGAN 



This work was supported by National Natural Science Foundation of China (Grant No. 61501251, 61373137, 61071167) and the Science Foundation of Nanjing University of Posts and Telecommunications Grant (NY214191).


  1. 1.
    Liu M-Y, Breuel T, Kautz J. Unsupervised image-to-image translation networks. CoRR, abs/1703.00848, 2017.Google Scholar
  2. 2.
    Hertzmann A, Jacobs CE, Oliver N, Curless B, Salesin DH. Image analogies. In: Proceedings of the 28th annual conference on computer graphics and interactive techniques, SIGGRAPH ’01. ACM: New York, NY, USA, 2001. p. 327–40.Google Scholar
  3. 3.
    Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A, Bengio Y. Generative adversarial nets. In: Ghahramani Z, Welling M, Cortes C, Lawrence ND, Weinberger KQ, editors. Advances in neural information processing systems 27. New York: Curran Associates, Inc.; 2014. p. 2672–80.Google Scholar
  4. 4.
    Perarnau G, van de Weijer J, Raducanu B, Álvarez JM. Invertible conditional gans for image editing. CoRR, abs/1611.06355, 2016.Google Scholar
  5. 5.
    Zhu J-Y, Krähenbühl P, Shechtman E, Efros AA. Generative visual manipulation on the natural image manifold. In: Proceedings of European conference on computer vision (ECCV), 2016.CrossRefGoogle Scholar
  6. 6.
    Xia Y, He D, Qin T, Wang L, Yu N, Liu T-Y, Ma W-Y. Dual learning for machine translation. CoRR, abs/1611.00179, 2016.Google Scholar
  7. 7.
    Shen W, Liu R. Learning residual images for face attribute manipulation. CoRR, abs/1612.05363, 2016.Google Scholar
  8. 8.
    Kim T, Cha M, Kim H, Lee JK, Kim J. Learning to discover cross-domain relations with generative adversarial networks. CoRR, abs/1703.05192, 2017.Google Scholar
  9. 9.
    Yi Z, Zhang H, Tan P, Gong M. Dualgan: Unsupervised dual learning for image-to-image translation. CoRR, abs/1704.02510, 2017.Google Scholar
  10. 10.
    Zhu J-Y, Park T, Isola P, Efros AA. Unpaired image-to-image translation using cycle-consistent adversarial networks. CoRR, abs/1703.10593, 2017.Google Scholar
  11. 11.
    Isola P, Zhu J-Y, Zhou T, Efros AA. Image-to-image translation with conditional adversarial networks. CoRR, abs/1611.07004, 2016.Google Scholar
  12. 12.
    Choi Y, Choi M-J, Kim M, Ha J-W, Kim S, Choo J. Stargan: Unified generative adversarial networks for multi-domain image-to-image translation. CoRR, abs/1711.09020, 2017.Google Scholar
  13. 13.
    Zhou S, Xiao T, Yang Y, Feng D, He Q, He W. Genegan: Learning object transfiguration and attribute subspace from unpaired data. CoRR, abs/1705.04932, 2017.Google Scholar
  14. 14.
    Taigman Y, Polyak A, Wolf L. Unsupervised cross-domain image generation. CoRR, abs/1611.02200, 2016.Google Scholar
  15. 15.
    Radford A, Metz L, Chintala S. Unsupervised representation learning with deep convolutional generative adversarial networks. CoRR, abs/1511.06434, 2015.Google Scholar
  16. 16.
    Qi G-J. Loss-sensitive generative adversarial networks on lipschitz densities. CoRR, abs/1701.06264, 2017.Google Scholar
  17. 17.
    Gulrajani I, Ahmed F, Arjovsky M, Dumoulin V, Courville AC. Improved training of wasserstein gans. CoRR, abs/1704.00028, 2017.Google Scholar
  18. 18.
    Heusel M, Ramsauer H, Unterthiner T, Nessler B, Klambauer G, Hochreiter S. Gans trained by a two time-scale update rule converge to a nash equilibrium. CoRR, abs/1706.08500, 2017.Google Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2020

Authors and Affiliations

  • Chen Wu
    • 1
    Email author
  • Lei Li
    • 1
  • Zhenzhen Yang
    • 1
  • Peihong Yan
    • 1
  • Jiali Jiao
    • 1
  1. 1.National Engineering Research Center of Communication and Network Technology School of ScienceNanjing University of Posts and TelecommunicationsNanjingChina

Personalised recommendations