Official repository for the paper DCFace: Synthetic Face Generation with Dual Condition Diffusion Model (CVPR 2023).

• Arxiv: https://arxiv.org/abs/2304.07060
• Main paper: main.pdf
• Supplementary: supp.pdf

1. one-liner : install.sh

If above fails, then

pip install -r requirements.txt
# and download model weights from the link below

We provide the sample code to generate images with the pretrained weights. The sample aligned images are provided in the repository.

• Download the pretrained weights from the link
• Place the pretrained_models directory under dcface (same level as src)
• Run

cd dcface/src
python synthesis.py --id_images_root sample_images/id_images/sample_57.png --style_images_root sample_images/style_images/woman

One can also generate new subject images and prepare custom style images.

Unconditional ID image generation is done in dcface/stage1/unconditional_generation Take a look at the README.md in that directory for instructions on how to generate new ID images.

Any aligned images can serve as style images. We provive some sample images in sample_images/style_images directory. For anyone who wants to use their own style images, one should align the images first. Take a look at the README.md in dcface/stage1/style_bank directory for instructions on how to align images.

Assuming that you followed 1. and 2. you will have an id_image and style_images directory. For the sake of explaination, let's say

• ID image is <Project_root>/dcface/stage1/unconditional_generation/unconditional_samples_aligned/00011.png
• Style directory is <Project_root>/dcface/stage1/style_bank/style_images/raw_aligned Then to combine these run by pointing at these paths,

cd dcface/src
python synthesis.py \
        --id_images_root <Project_root>/dcface/stage1/unconditional_generation/unconditional_samples_aligned/00011.png \
        --style_images_root <Project_root>/dcface/stage1/style_bank/style_images/raw_aligned

The result will be saved at <Project_root>/dcface/generated_images/

• Download casia webface dataset from insightface
  • Place it under $DATA_ROOT (ex: /data/).
  • ex: /data/faces_webface_112x112

• Download all pretrained weights from the link
• Place the pretrained_models directory under dcface (same level as src)

cd dcface/src/
bash train.sh

DCFace synthetic dataset can be downloaded from link

• dcface_0.5m_oversample_xid.zip
• dcface_1.2m_oversample_xid.zip

The format of the downloaded file is in rec format.

• you can convert it to png using the script.
• rec file will be useful for the face recognition training script provided in the repository. (to be released soon)

cd dcface/convert
python record.py --rec_path <path_to_rec_file> --save_path <path_to_save_png>
# ex
# <path_to_rec_file> : dcface_0.5m_oversample_xid/record
# <path_to_save_png> : dcface_0.5m_oversample_xid/images

• to be released soon