HandR2N2: Iterative 3D Hand Pose Estimation Using a Residual Recurrent Neural Network

Wencan Cheng and Jong Hwan Ko

IEEE International Conference on Computer Vision (ICCV), 2023

---

1. Prepare dataset

   please download the NYU dataset

   follow the instructions in the './preprocess_nyu/' for datasets preprocessing

2. Install PointNet++ CUDA operations

   follow the instructions in the './pointnet2' for installation

3. Evaluate

   execute python3 eval_rrnn.py --model [saved model name] --iters [training iterations] --test_iters [testing iterations] --test_path [testing set path]

   for example python3 eval_rrnn.py --model best_model.pth --iters 3 --test_iters 5 --test_path ../preprocess_nyu/testing/

   we provided the pre-trained models ('./pretrained_model/nyu_rrnn_3iters/best_model.pth') for NYU

4. If a new training process is needed, please execute the following instructions after step 1 and 2 are completed

   . for training NYU execute python3 train_rrnn.py --iters [number of training iteration] --dataset_path [NYU training dataset path] --test_path [NYU testing dataset path] example python3 train_rrnn.py --iters 3 --dataset_path ../preprocess_nyu/training/ --test_path ../preprocess_nyu/testing/