Skip to content
/ Realistic-Image-Generation-from-Sketches-Using-Conditional-GANs Public

Developed a conditional GAN to generate realistic images from unpaired sketches, conditioned on class labels, and evaluated performance using FID and IS metrics. Conducted experiments to compare generated image accuracy with original dataset accuracy and integrated WandB for training logs and visualizations.

0 stars 0 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

Manaswi-Vichare/Realistic-Image-Generation-from-Sketches-Using-Conditional-GANs

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
CGAN.ipynb
CGAN.ipynb
 
 
README.md
README.md
 
 

Repository files navigation

Realistic-Image-Generation-from-Sketches-Using-Conditional-GANs

Objective:

  1. To develop a system to generate realistic images from sketches using conditional Generative Adversarial Network (GAN), where the generated images are based on the label provided as the input.
  2. The generated images should represent the characteristics associated with the label given as input.

Dataset: ISIC Dataset.
Images paired with their respective sketch images. Visualization of the dataset can be observed below:
image

Generated images on Test Data after 50 Epochs:
Note: These images were generated after only 50 epochs. Better results can be obtained by training the model for ideally 2000+ epochs.
image

About

Developed a conditional GAN to generate realistic images from unpaired sketches, conditioned on class labels, and evaluated performance using FID and IS metrics. Conducted experiments to compare generated image accuracy with original dataset accuracy and integrated WandB for training logs and visualizations.

Resources

Readme
Activity

Stars

0 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages