3DGlobalFormer

3DGlobalFormer: Three Domain Global Feature Fusion in 3D Human Estimation

Thank you for your interest, the code and checkpoints are being updated.

The released codes include

checkpoint/:                        the folder for model weights of 3DGlobalFormer.
dataset/:                           the folder for data loader.
common/:                            the folder for basic functions.
model/:                             the folder for 3DGlobalFormer network.
run_global.py:                        the python code for 3DGlobalFormer networks training.

Environment

Make sure you have the following dependencies installed:

• PyTorch >= 0.4.0
• NumPy
• Matplotlib=3.1.0

Datasets

Our model is evaluated on Human3.6M and MPI-INF-3DHP datasets.

Human3.6M

We set up the Human3.6M dataset in the same way as VideoPose3D.

MPI-INF-3DHP

We set up the MPI-INF-3DHP dataset in the same way as P-STMO.

Evaluation

You can download our pre-trained models from Google Drive. Put them in the ./checkpoint directory.

Human3.6M

To evaluate our 3DGlobalFormer model on the 2D keypoints obtained by CPN, please run:

 python run_global.py -f 243 -b 128 --train 0 --layers 6 -s 1 -k 'cpn_ft_h36m_dbb' --reload 1 --previous_dir ./checkpoint/your_best_epoch.pth