Awesome

Multi-Context Attention for Human Pose Estimation

This repository includes Torch code for evaluation and training of the network presented in:

Xiao Chu, Wei Yang, Wanli Ouyang, Cheng Ma, Alan L. Yuille, Xiaogang Wang, Multi-Context Attention for Human Pose Estimation, CVPR, 2017. (arXiv preprint)

The code is developed upon Stacked Hourglass Network.

Installation

To run this code, the following packages must be installed:

• Torch7
• hdf5 (and the torch-hdf5 package)
• cudnn
• qlua (for displaying results)
• matio: to save predictions in Matlab's .mat file.

Testing

For testing, please go to the test directory and follow the README for instructions.

Training

For training, please go to the train directory and follow the README for instructions.