HandFoldingNet ✌️ : A 3D Hand Pose Estimation Network Using Multiscale-Feature Guided Folding of a 2D Hand Skeleton

Wencan Cheng, Jae Hyun Park, Jong Hwan Ko

IEEE International Conference on Computer Vision (ICCV), 2021\

arXiv preprint:https://arxiv.org/abs/2108.05545

1. Prepare dataset

   please download the ICVL and MSRA dataset, and put them under path './data/ICVL/' and './data/MSRA/', respectively.

   execute instructions in the './preprocess_icvl/' and './preprocess_msra/' for datasets preprocessing

2. Evaluate

   navigate to "./train_eval" directory

   execute python3 eval_[dataset name]_folding.py --model [saved model name] --test_path [testing set path]

   for example on ICVL

   python3 eval_icvl_folding.py --model netR_SOTA.pth --test_path ../data/ICVL_center_pre0/Testing/

   or on MSRA

   python3 eval_msra_folding.py --model netR.pth --test_path ../data/msra_preprocess/

   we provided the pre-trained models ('./results/icvlfolding/netR_SOTA.pth' and './results/msrafolding/P0/netR.pth') for testing ICVL and MSRA

   we also provided the predicted labels located at './labels' directory for visualizing the performance through awesome-hand-pose-estimation

3. If a new training process is needed, please execute the following instructions after step1 is completed

   navigate to "./train_eval" directory

   . for training MSRA

   execute python3 train_msra_folding.py --dataset_path [MSAR dataset path]

   example python3 train_msra_folding.py --dataset_path ../data/msra_preprocess/

   . for training ICVL

   execute python3 train_icvl_folding.py --train_path [ICVL training dataset path] --test_path [ICVL testing dataset path]

If you find our code useful for your research, please cite our paper

@inproceedings{cheng2021handfoldingnet,
  title={HandFoldingNet: A 3D Hand Pose Estimation Network Using Multiscale-Feature Guided Folding of a 2D Hand Skeleton},
  author={Cheng, Wencan and Park, Jae Hyun and Ko, Jong Hwan},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={11260--11269},
  year={2021}
}