Aircap_Pose_Estimator

Get this repo: git clone https://github.com/robot-perception-group/Aircap_Pose_Estimator.git

Data

• Go to https://aircapdata.is.tue.mpg.de, register and login.
• Download AirCap-Pose_Estimator-minimaldata from https://aircapdata.is.tue.mpg.de/downloads
• Extract the contents (directory named "data") to the repo dir.

Install Requirements

• Python2.7

• ROS Melodic [http://wiki.ros.org/melodic]

• Other requirementas pip install -r requirements.txt

• torchgeometry v0.1.0

  it clone https://github.com/arraiyopensource/kornia.git
  d kornia
  it checkout v0.1.0
  ython setup.py install
  

• Download SMPL from [http://smpl.is.tue.mpg.de/] and extract its content in the parent directory i.e. Aircap_Pose_Estimator/

  optional requirements

  • Mayavi for results visualization. Install Mayavi for Python2.7 from https://docs.enthought.com/mayavi/mayavi/installation.html#installing-with-pip

Run Aircap_Pose_Estimator demo

• run Aircap_Pose_Estimator demo as in paper python fittingscript.py /path/to/result/directory
• results will be saved in /path/to/result/directory. Mean error for each joint will be in the file /path/to/result/directory/final_err_res.npy
• To visualize results, execute python viz_res.py /path/to/result/directory to launch the visualization of results alongwith the ground truth.