GAIP

Source code of Graph-Based Adversarial Inertial Poser (GAIP).

Usage

Dependencies

We use python 3.7.6. You can install the package as requirements.py.

SMPL body model

Download SMPL model from here. You should click SMPL for Python and download the version 1.0.0 for Python 2.7 (10 shape PCs). Then unzip it.

Pre-trained network weights

Download weights from here. A better version is updated (2024.3.28).

Datasets

1. Public dataset: dataset AMASS, DIP-IMU and TotalCapture can be obtained following the instructions from Transpose.
2. SingleOne-IMU dataset is available from here.
3. Miaxmo-IMU dataset is available from here. Detailed descriptions about our own dataset will be sorted out soon.

Run the evaluation

Directly download .npy files of datasets, then run eval.py in VS Code. Robust project execution process is being sorted out. According to the different dataset path set in dataset_eval.py, different results will be printed.