GAN Theory
Modifyingthe Optimization of GAN
| 題目 |
內容 |
| GAN |
|
| DCGAN |
|
| WGAN |
|
| Least-square GAN |
|
| Loss Sensitive GAN |
|
| Energy-based GAN |
|
| Boundary-seeking GAN |
|
| Unroll GAN |
Different Structure from the Original GAN
| 題目 |
內容 |
| Conditional GAN |
|
| Semi-supervised GAN |
|
| InfoGAN |
|
| BiGAN |
|
| Cycle GAN |
|
| Disco GAN |
|
| VAE-GAN |
|
| LAPGAN |
用了多個GAN可生成高分辨率圖像 |
GAN Application
| pix2pix |
|
| 題目 |
內容 |
| Image-to-Image Translation with Conditional Adversarial Networks |
image2image、paired Image-to-Image Translation |
| High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs |
image2imageHD、paired Image-to-Image Translation |
| CycleGAN |
Unpaired Image-to-Image Translation |
| Disco GAN |
側重分析雙向映射，或者說 bijective mapping 的約束：避免 mode collapse 進而提升生成樣本質量的 |
| DualGAN |
生成器和判別器都和pix2pix一樣。 用了wgan來訓練。 |
| 注：最后三篇論文的想法十分相似，幾乎可以說是孿生三兄弟 |
|
| text2image |
|
| 題目 |
內容 |
| 人臉生成 |
|
| 題目 |
內容 |
| Face-generator - Generate human faces with neural networks |
|
| Beyond Face Rotation: Global and Local Perception GAN for Photorealistic and Identity Preserving Frontal View Synthesis |
根據單一側臉生成正面逼真人臉 |
| NEURAL FACE |
use DCGAN、鏈接：https://carpedm20.github.io/faces/ |
| 注：DCGAN、WGAN這類都可以生成人臉 |
|
| 按生成的圖片種類分 |
|
| 題目 |
內容 |
| 生成卧室 |
DCGAN、WGAN |
| 生成動漫頭像 |
DCGAN |