How to use Action Recognition Python* Demo for Ubuntu

[TNaka1]

Dear Intel Support

We want to check the Action Recognition Python* Demo for Ubuntu Open VINO.

But we don't understand to how to use this demo. Please give us advice.

Best Regards,

Tasuku Nakajima 

[Shubha_R_Intel]

Dear Tasuku N,

Hopefully the below online doc should help you:

https://docs.openvinotoolkit.org/latest/_inference_engine_ie_bridges_python_sample_action_recognition_README.html

You can download the required models from here:

https://download.01.org/opencv/

Thanks,

Shubha

[TNaka1]

Dear Shubha-san

Thank you for your support.

Do you have Media Files Available for Samples to run the Action Recognition sample applications?

We couldn't find it on https://github.com/intel-iot-devkit/sample-videos.

Best Regards,

Tasuku Nakajima

[Shubha_R_Intel]

Dear Tasuku N,

You should be able to download videos from the website below:

https://deepmind.com/research/open-source/open-source-datasets/kinetics/

Thanks,

Shubha

[Vadim_A_Intel]

Dear Tasuku N,

There are two pre-trained models for action recognition in the open model zoo.

The first is action-recognition-0001 (encoder + decoder) which is general-purpose human action recognition model, trained on Kinetics dataset, hence you can use videos from Kinetics dataset, as Shubha R suggested. Also, there is action recognition model for driver monitoring scenario - driver-action-recognition-adas-0002 (encoder + decoder), we submitted a PR with sample video for this scenario (it is not merged yet, but you can use this link: https://github.com/vadimadr/sample-videos/raw/va/add_action_recognition_sample/driver-action-recognition.mp4 to download the video)

Vadim