Conditional GAN Floorplans.
A PyTorch research implementation of text-conditioned floorplan image synthesis, with model code, saved weights, and sample outputs.
RESEARCH DEEP DIVE
The research approach.
Text-conditioned image generation can explore visual layout variations from a room description. This study implements that idea with a conditional generative adversarial network.
PyTorch generator and discriminator modules combine a text encoding with latent input. The repository includes an inference CLI, a saved generator checkpoint, generated image samples, and FID/IS evaluation code.
Explore the model modules, saved weights, sample images, and evaluation code. The study focuses on exploratory image synthesis conditioned by room-description encodings.
FROM THE REPOSITORY
What's inside.
- 01
Combines a room-description encoding with latent input for conditional generation.
- 02
Includes generator and discriminator modules, an inference CLI, and a saved generator checkpoint.
- 03
Provides generated image variations and FID/IS evaluation code for further investigation.
These notes summarize the reviewed implementation and available artifacts. Open the original source for code, documentation, and subsequent changes.
Open the original repository