PyTorch implementation of 3D-Action-recognition

PyTorch implementation of Skeleton-based Action Recognition With Convolutional Neural, IEEE 2017.

Accuracy(Cross-View): 87.6%（The baseline is 89.3%）

Network framework

Requirements

You need to install the following packages.

• numpy
• torch-1.0.1
• torchvision-0.3.0
• hdf5

Dataset download

• Official website application download
• BaiduNetdisk Code：dgoi
• BaiduNetdisk Code：7vlp

hdf5 data(only cross-view)

• BaiduNetdisk Code：t2au

Tips

If you have any questions about my work, please do not hesitate to contact me by emails. [email protected]