- Compare conditional GAN with the basic GAN.
- Understanding of adversarial networks.
- Loss function without hand-engineering.
- Generative model to come up with a way of matching their generated distribution to a real data distribution, i.e., Minimizing the distance between the two distributions is critical for creating a system that generates content that looks good, new, and like it is from the original data distribution.
Dataset used : facades data
Link to the dataset : http://efrosgans.eecs.berkeley.edu/pix2pix/datasets/
The results in this experiment suggest that conditional adversarial networks are a promising approach for many image-image translation tasks, especially those involving highly structured graphical outputs. These networks learn a loss adapted to the task and data at hand, which makes them applicable in a wide variety of settings.
Phillipi Isola Thanks for the wonderful paper on Image-Image translation using Conditional Adversarial Network