Image-to-Image Translation with Conditional Adversarial Networks
Citations
5,354 citations
4,465 citations
Cites background or methods from "Image-to-Image Translation with Con..."
...[20], which uses a conditional generative adversarial network [14] to learn a mapping from input to output images....
[...]
...Such a patch-level discriminator architecture has fewer parameters than a fullimage discriminator, and can be applied to arbitrarily-sized images in a fully convolutional fashion [20]....
[...]
...From left to right: input, BiGAN/ALI [6, 7], CoGAN [28], CycleGAN (ours), pix2pix [20] trained on paired data, and ground truth....
[...]
...Using the same evaluation datasets and metrics as “pix2pix” [20], we compare our method against several baselines both qualitatively and quantitatively....
[...]
...For the discriminator networks we use 70×70 PatchGANs [20, 26, 25], which aim to classify whether 70 × 70 overlapping image patches are real or fake....
[...]
3,838 citations
Cites methods from "Image-to-Image Translation with Con..."
...Finally, modern methods for image synthesis train deep networks with patch-based losses (implemented as convolutions) [8, 21]....
[...]
3,322 citations
3,227 citations
References
111,197 citations
"Image-to-Image Translation with Con..." refers methods in this paper
...We use minibatch SGD and apply the Adam solver [29]....
[...]
49,590 citations
40,609 citations
38,211 citations
"Image-to-Image Translation with Con..." refers background or methods in this paper
...Fortunately, this is exactly what is done by the recently proposed Generative Adversarial Networks (GANs) [22, 12, 41, 49, 59]....
[...]
...Figure 1 of the original GAN paper [22])....
[...]
...An adversarial loss, on the other hand, can in principle become aware that grayish outputs are unrealistic, and encourage matching the true color distribution [22]....
[...]
...To optimize our networks, we follow the standard approach from [22]: we alternate between one gradient descent step on D, then one step on G....
[...]
...GANs are generative models that learn a mapping from random noise vector z to output image y, G : z → y [22]....
[...]
30,811 citations