This is the official implementation for the paper, "Realistic Depth Image Synthesis for 3D Hand Pose Estimation".
@ARTICLE{10310128,
author={Zhou, Jun and Xu, Chi and Ge, Yuting and Cheng, Li},
journal={IEEE Transactions on Multimedia},
title={Realistic Depth Image Synthesis for 3D Hand Pose Estimation},
year={2023},
volume={},
number={},
pages={1-12},
doi={10.1109/TMM.2023.3330522}}
conda env create -f env_Sty2.yaml
Download the checkpoint for nyu style image at here.
Download the noise-free image at here.
More noise-free image can be synthesized following this work.
Finally, the directories look like
├── ckpt
│ └── nyu_style.pkl
├── data
│ ├── 2-0_Is
│ └── mean_std.npy
├── Dataset_utils
├── Dnnlib
├── env_Sty2.yaml
├── generate_nyu.py
├── Metrics
├── Networks
└── Torch_utils
python generate_nyu.py
We thanks HANDS19-Challenge-Toolbox for providing a simple way to synthesize noise-free images.
We thank the following repos providing helpful components/functions in our work.