Makeup Transfer with GAN

Overview

This project aims to reproduce the CA-GAN paper using our own dataset, which includes a variety of skin tones based on the Monk Skin Tone Scale. The goal is to generate predicted lipstick colors based on the skin tone and the original images provided.

Dependencies

• Python 3.5+
• PyTorch 0.4.0+
• TensorFlow 1.3+ (optional for TensorBoard)

Steps

Step 1: Retrieve Original and Target Lip Makeup Color

• Original Lip Color: Crop the lips from the image to extract the makeup color.
  • Example:

  

• Target Makeup Color: Display the result of the retrieved makeup color on the face.
  • Example:

  

Step 2: Train with CA-GAN Model

• Train the model using the CA-GAN architecture.
• Model Structure:
  • Example: jpg/model.png

Step 3: Generate Results

• Generate and visualize the results.
• Display charts and final images.

Results

• Makeup Transfer: Left: source image / Middle: result image / Right: target image.

  • Example:
  

• Dark Skin Tone: Left: source image / Middle: result image / Right: target image.

  • Example:
  

• Light Skin Tone: Left: source image / Middle: result image / Right: target image.

  • Example:
  

Citation

• CA-GAN Paper: Link to paper
• StarGAN Paper: Link to paper

@inproceedings{choi2018stargan,
  author={Yunjey Choi and Minje Choi and Munyoung Kim and Jung-Woo Ha and Sunghun Kim and Jaegul Choo},
  title={StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  year={2018}
}

@inproceedings{kips2020gan,
  title={Ca-gan: Weakly supervised color aware gan for controllable makeup transfer},
  author={Kips, Robin and Gori, Pietro and Perrot, Matthieu and Bloch, Isabelle},
  booktitle={Computer Vision--ECCV 2020 Workshops: Glasgow, UK, August 23--28, 2020, Proceedings, Part III 16},
  pages={280--296},
  year={2020},
  organization={Springer}
}