KD-DLGAN

KD-DLGAN: Data Limited Image Generation via Knowledge Distillation

Kaiwen Cui, Yingchen Yu, Fangneng Zhan, Shengcai Liao, Shijian Lu, Eric Xing

CVPR 2023

Samples from the style-ganv2 - trained 680 epochs

DA (Baseline)

KD-DLGAN

This folder provides a re-implementation of this paper in PyTorch, developed as part of the course METU CENG 796 - Deep Generative Models. The re-implementation is provided by:

• Ibrahim Ersel Yiğit, [email protected]
• Meriç Karadayı, [email protected]

Please see the jupyter notebook file main.ipynb for a summary of paper, the implementation notes and our experimental results.

Additional:

• This folder is implemented based on the DiffAugment for StyleGAN2 (Pytorch) with minimal modifications for training. The Jupyter notebook file main.ipynb first clones the original DiffAugment for StyleGAN2 (Pytorch) repository then, overwrites the original files with the modified networks.py and train.py files and adds the kd_loss.py to the cloned repository.