Domain Disentangled Generative Adversarial Network for Zero-Shot Sketch-Based 3D Shape Retrieval

by Rui Xu, Zongyan Han, Le Hui, Jianjun Qian, and Jin Xie.

Usage

1. requires:

   CUDA10 + Pytorch 1.2.0 + Python3
   

2. Train:

   CUDA_VISIBLE_DEVICES=0 python main.py --network DD_GAN --model_dir DD_GAN  --batch_size 256 --max_epoch 300  --snapshot 50 --phase train
   

3. Test:

   CUDA_VISIBLE_DEVICES=0 python main.py --network DD_GAN --model_dir DD_GAN  --batch_size 20  --pretrain_model best_full_P.pth --phase test
   

Citation

If you find the code useful, please consider citing:

@article{xu2022domain,
  title={Domain Disentangled Generative Adversarial Network for Zero-Shot Sketch-Based 3D Shape Retrieval},
  author={Xu, Rui and Han, Zongyan and Hui, Le and Qian, Jianjun and Xie, Jin},
  journal={arXiv preprint arXiv:2202.11948},
  year={2022}
}

Acknowledgement

Our preprocessing dataset

Our word embedding model is from GloVe

Our evaluation code is from Deep Correlated Holistic Metric Learning for Sketch-based 3D Shape Retrieval