Synthetic Image Generation with GANs
Explored Generative Adversarial Networks for synthetic image generation using the classic generator-discriminator training setup.
Project Snapshot
| Signal | Details |
|---|---|
| Timeline | Feb 2018 to Jun 2018 |
| Organization | Xiamen University |
| Focus | Generating synthetic images using Generative Adversarial Networks |
| Stack | Custom implementation and project-specific tooling |
Model Structure
| Component | Role |
|---|---|
| Generator | Converts random noise into synthetic images |
| Discriminator | Classifies images as real or generated |
| Training goal | Make generated images realistic enough to fool the discriminator |
Key Idea
The discriminator learns to separate real training images from fake images, while the generator learns to produce increasingly realistic images through this adversarial feedback loop.