Code for CT-Net: Complementary Transfering Network for Garment Transfer with Arbitrary Geometric Changes，CVPR'21.

[Paper] [Supp]

Garment transfer shows great potential in realistic applications with the goal of transfering outfits across different people images. However, garment transfer between images with heavy misalignments or severe occlusions still remains as a challenge. In this work, we propose Complementary Transfering Network (CT-Net) to adaptively model different levels of geometric changes and transfer outfits between different people. In specific, CT-Net consists of three modules: i) A complementary warping module first estimates two complementary warpings to transfer the desired clothes in different granularities. ii) A layout prediction module is proposed to predict the target layout, which guides the preservation or generation of the body parts in the synthesized images. iii) A dynamic fusion module adaptively combines the advantages of the complementary warpings to render the garment transfer results. Extensive experiments conducted on DeepFashion dataset demonstrate that our network synthesizes high-quality garment transfer images and significantly outperforms the state-of-art methods both qualitatively and quantitatively.

Clone the Synchronized-BatchNorm-PyTorch repository.

cd models/networks/
git clone https://github.com/vacancy/Synchronized-BatchNorm-PyTorch
cp -rf Synchronized-BatchNorm-PyTorch/sync_batchnorm .
cd ../../

Install dependencies:

pip install -r requirements.txt

• Download the In-shop Clothes Retrieval Benchmark of DeepFashion.We use images of resolution 256 * 256. Save it as dataset/img.
• We use OpenPose to estimate pose of DeepFashion. Download the openpose results and save it as dataset/pose.
• We use LIP_JPPNet and Densepose to estimate the segmentation and densepose descriptor of DeepFashion. Download the seg and densepose results and save it as dataset/seg_dp.
• We use Irregular Mask Dataset to randomly remove part of the limbs in the body images in the dynamic generation module for mimicking image inpainting. Downloads the dataset and save it as dataset/train_mask.

Download the pretrained model from here and save them in 'checkpoints/one_corr_tps'

Then run the command.

python test.py 

Download the pretrained VGG weights for vgg-based loss from here and save it at model/. Our model is trained in two stages.

• First, the Complementary Warping Module is trained for 20 epoches to estimate reasonable warpings. Run the command.

  python -m visdom.server -port 8097
  python train.py --niter 20 --niter_decay 0 --display_port 8097 --train_corr_only
  

• Then our model is jointly trained in an end-to-end manner for another 80 epoches. Learning rate is fixed at 0.0002 for the first 40 epoches and then decays to zero linearly in the remaining steps. Run the command.

  python train.py --niter 40 --niter_decay 40 --display_port 8097
  

If you use this code for your research, please cite our papers.

@inproceedings{yang2021ct,
  title={CT-Net: Complementary Transfering Network for Garment Transfer with Arbitrary Geometric Changes},
  author={Yang, Fan and Lin, Guosheng},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={9899--9908},
  year={2021}
}

The use of this software is RESTRICTED to non-commercial research and educational purposes.

This code borrows heavily from CoCosNet. We also thanks Jiayuan Mao for his Synchronized Batch Normalization code.