Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
Paper
•
1703.10593
•
Published
•
2
This model is based on CycleGAN architecture. It takes images, and generates a futuristic neon image for the image provided.Hope this model neonifies your images.

The model is trained on 256x256 high contrasted neon images as style images, and normal images (including people,scenery etc.) as base images.
All details regarding how to use the model, fine-tune it, are added to GitHub.
Check out the spaces demo, and try the model by yourselves.
Hope you all enjoy it!