English

Cycle In Cycle Generative Adversarial Networks for Keypoint-Guided Image Generation

Computer Vision and Pattern Recognition 2020-04-17 v3 Machine Learning Image and Video Processing

Abstract

In this work, we propose a novel Cycle In Cycle Generative Adversarial Network (C2^2GAN) for the task of keypoint-guided image generation. The proposed C2^2GAN is a cross-modal framework exploring a joint exploitation of the keypoint and the image data in an interactive manner. C2^2GAN contains two different types of generators, i.e., keypoint-oriented generator and image-oriented generator. Both of them are mutually connected in an end-to-end learnable fashion and explicitly form three cycled sub-networks, i.e., one image generation cycle and two keypoint generation cycles. Each cycle not only aims at reconstructing the input domain, and also produces useful output involving in the generation of another cycle. By so doing, the cycles constrain each other implicitly, which provides complementary information from the two different modalities and brings extra supervision across cycles, thus facilitating more robust optimization of the whole network. Extensive experimental results on two publicly available datasets, i.e., Radboud Faces and Market-1501, demonstrate that our approach is effective to generate more photo-realistic images compared with state-of-the-art models.

Keywords

Cite

@article{arxiv.1908.00999,
  title  = {Cycle In Cycle Generative Adversarial Networks for Keypoint-Guided Image Generation},
  author = {Hao Tang and Dan Xu and Gaowen Liu and Wei Wang and Nicu Sebe and Yan Yan},
  journal= {arXiv preprint arXiv:1908.00999},
  year   = {2020}
}

Comments

9 pages, 8 figures, accepted to ACM MM 2019

R2 v1 2026-06-23T10:38:32.265Z