Learning Human Poses from Actions

This repository is the PyTorch implementation for the network presented in:

Aditya Arun, C.V. Jawahar, and M. Pawan Kumar,
Learning Human Poses from Actions,
[BMVC 2018](arXiv:1807.09075, 2018).

[Project Page] [Paper] [Supplementary] [ArXiv]

If you use this code, please consider citing our work.

@inproceedings{posesfromactions2018,
  title={Learning Human Poses from Actions},
  author={Arun, Aditya and Jawahar, C.~V and Kumar, M. Pawan},
  booktitle={BMVC},
  year={2018}
}

Requirements

cudnn
PyTorch
Python with h5py, opencv and progress
Optional: tensorboard

Training

Modify src/ref.py to setup the MPII dataset path and root directory.
Run Prediction Net

python main.py -arch pred_net

Run Conditional Net

python main.py -arch cond_net

adityaarun1/human-poses-from-actions

Learning Human Poses from Actions

Requirements

Training