va-cnn

View Adaptive Neural Networks (VA) for Skeleton-based Human Action Recognition

Input

(Skeleton data from NTU-RGB+D dataset https://github.com/shahroudy/NTURGB-D/)

Output

Action estimate

Action estimate -> drink water

Usage

For this model use the skeleton data of NTU-RGB+D dataset.
Download the NTU-RGB+D dataset to try out the sample code.

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample data,

$ python3 va-cnn.py

If you want to specify the input skeleton file, put the file path after the --input option.

$ python3 va-cnn.py --input FILE_PATH

You can use --save_video option to output a video of the input skeleton data.

$ python3 va-cnn.py --save_video

Reference

View Adaptive Neural Networks (VA) for Skeleton-based Human Action Recognition

Framework

Pytorch

Model Format

ONNX opset=11

Netron

va-cnn.onnx.prototxt