Few-Shot Image Generation by
Conditional Relaxing Diffusion Inversion

Yu Cao^* ,  Shaogang Gong^†

Queen Mary University of London

ECCV 2024, Sun Sep 29th through Fri Oct 4th, 2024 at MiCo Milano.

[Project Page]   [Paper]  

Usage

1. Download Pre-trained Model

Download FFHQ 256*256 ckpt of Guided Diffusion from ddpm-segmentation to checkpoints/ddpm/ffhq.pt

2. Update Directory Paths

Modify the path in following files:

./datasets/babies_target/babies.csv
./scripts/fs_gradient_evaluate.py

3. Run Experiments

In the file main.sh, we provide commands to reproduce experiments results on Babies.

bash main.sh

Bibtex

If you find this project useful in your research, please consider citing our paper:

@article{cao2024few,
  title={Few-Shot Image Generation by Conditional Relaxing Diffusion Inversion},
  author={Cao, Yu and Gong, Shaogang},
  journal={arXiv preprint arXiv:2407.07249},
  year={2024}
}

Acknowledgments

Parts of this project page were adopted from the Nerfies page.

Website License

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.