Conditional GAN

Posted on | In

  1. Conditioned on label vector: conditional GAN [4], CVAE-GAN [6]

  2. Conditioned on a single image

    • pix2pix [1]; high-resolution pix2pix [2] (add coarse-to-fine strategy); BicycleGAN [3] (combination of cVAE-GAN and cLR-GAN)
    • DAGAN [5]

Reference

[1] Isola, Phillip, et al. “Image-to-image translation with conditional adversarial networks.” CVPR, 2017

[2] Wang, Ting-Chun, et al. “High-resolution image synthesis and semantic manipulation with conditional gans.” CVPR, 2018.

[3] Zhu, Jun-Yan, et al. “Toward multimodal image-to-image translation.” NIPS, 2017.

[4] Mirza, Mehdi, and Simon Osindero. “Conditional generative adversarial nets.” arXiv preprint arXiv:1411.1784 (2014).

[5] Antoniou, Antreas, Amos Storkey, and Harrison Edwards. “Data augmentation generative adversarial networks.” arXiv preprint arXiv:1711.04340 (2017).

[6] Bao, Jianmin, et al. “CVAE-GAN: fine-grained image generation through asymmetric training.” ICCV, 2017.