Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Fine-tuning StyleGAN2 for Cartoon Face Generation
| Date | Stars |
|---|---|
| 2026-07-31 | 656 |
| 2026-08-05 | 655 |
| 2026-08-06 | 655 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Cartoon-StyleGAN 🙃 : Fine-tuning StyleGAN2 for Cartoon Face Generation > **Abstract** > > Recent studies have shown remarkable success in the unsupervised image to image (I2I) translation. However, due to the imbalance in the data, learning joint distribution for various domains is still very challenging. Although existing models can generate realistic target images, it’s difficult to maintain the structure of the source image. In addition, training a generative model on large data in multiple domains requires a lot of time and computer resources. To address these limitations, I propose a novel image-to-image translation method that generates images of the target domain by finetuning a stylegan2 pretrained model. The stylegan2 model is suitable for unsupervised I2I translation on unbalanced datasets; it is highly stable, produces realistic images, and even learns properly from limited data when applied with simple fine-tuning techniques. Thus, in this project, I propose new methods to preserve the structure of the source images and generate realistic images in the target domain. <p align='center'><img src="https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/Result.gif?raw=1" width = '700'></p> **Inference Notebook** 🎉 You can do this task in colab ! : [](https://colab.research.google.com/github/happy-jihye/Cartoon-StyleGan2/blob/main/Cartoon_StyleGAN2.ipynb) **Arxiv** [](https://arxiv.org/abs/2106.12445) **[NEW!] 2021.08.30 Streamlit Ver** - [`cartoon-stylegan streamlit inference repo`](https://github.com/happy-jihye/Streamlit-Tutorial/tree/main/cartoon-stylegan) <p align='center'><img src="https://github.com/happy-jihye/Streamlit-Tutorial/blob/main/asset/cartoon-stylegan-1.gif?raw=1?raw=1" width = '700'></p> --- ## 1. Method ### Baseline : StyleGAN2-ADA + FreezeD <p align='center'><img src='https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/baseline.PNG?raw=1' width = '700' ></p> It generates realistic images, but does not maintain the structure of the source domain. | | | | ---- | ---- | | <img src="https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/baseline-rom101.gif?raw=1"> | <img src="https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/baseline-simpson.gif?raw=1"> | ### Ours : FreezeSG (Freeze Style vector and Generator) <p align='center'><img src='https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/latent.PNG?raw=1' width = '800' ></p> [FreezeG](https://github.com/bryandlee/FreezeG) is effective in maintaining the structure of the source image. As a result of various experiments, I found that not only the initial layer of the generator but also the initial layer of the style vector are important for maintaining the structure. Thus, I froze the low-resolution layer of both the generator and the style vector. **Freeze Style vector and Generator** <p align='center'><img src='https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/FreezeSG.PNG?raw=1' width = '800' ></p> **Results** <p align='center'><img src='https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/FreezeSG2.PNG?raw=1' width = '800' ></p> **With [Layer Swapping](https://arxiv.org/abs/2010.05334)** When LS is applied, the generated images by FreezeSG have a higher similarity to the source image than when FreezeG or the baseline (FreezeD + ADA) were used. However, since this fixes the weights of the low-resolution layer of the generator, it is difficult to obtain meaningful results when layer swapping on the low-resolution layer. <p align='center'><img src='https://github.com/happy-jihye/Cartoon-StyleGan2/blob/main/asset/images/FreezeSG3.png?raw=1' width = '800' ></p> ### Ours : Structure Loss
Excerpt of 15,499 characters
Read on GitHub
20
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:df27a86134fb241c, topic:gan
matched fp:df27a86134fb241c, desc:fine-tuning, desc:fine tuning