topic Re: Is there a framework/tool for continues arm and/or leg gesture recognition? in Items with no label

Is there a framework/tool for continues arm and/or leg gesture recognition?

BLiet — Thu, 18 Apr 2019 15:30:10 GMT

We have a D415 Camera. I just learn the basics at the moment. Out aim is to recognize some arm and leg gestures. Some years ago we did this with a Kinect v2.

Is there some framework or tool for this purpose?

Or how did someone something like this before and can help to getting started?

Re: Is there a framework/tool for continues arm and/or leg gesture recognition?

MartyG — Thu, 18 Apr 2019 19:47:23 GMT

The best solution available for skeletal tracking with the RealSense D415 or D435 camera is commercial software called Nuitrack SDK. It is priced at around $30 a year and has a free trial version.

https://nuitrack.com

Re: Is there a framework/tool for continues arm and/or leg gesture recognition?

BLiet — Thu, 18 Apr 2019 20:48:19 GMT

Thanks for the answer. I notice NuiTrack. But it just support some basic gestures (Push, Wave, Swipe) and has no possibility to add custom gestures.

So I think we have to analyse the Skeleton data fron nuitrack with an additional framework.

Re: Is there a framework/tool for continues arm and/or leg gesture recognition?

MartyG — Thu, 18 Apr 2019 22:29:51 GMT

Custom gestures can be designed with the older RealSense SR300 camera model using software called TouchDesigner and a component it has called CHOP. This was possible because the SR300 has support for hand joint and face landmark detection and tracking in the 2016 version of the RealSense SDK.

https://docs.derivative.ca/RealSense

https://www.youtube.com/watch?v=YFD7wDMP984

The CHOP component does not support the D415 or D435 cameras, unfortunately.

The SR300 camera is still available in Intel's Click online store if you would like to explore this possibility. The camera uses structured light patterns, like Kinect V1 does.

https://store.intelrealsense.com/buy-intel-realsense-developer-kit-sr300.html

If you would like to explore the capture and harnessing of tracking data with the D415, Intel's presentation in the YouTube link below may be of interest to you.

https://www.youtube.com/watch?v=VSHDyUXSNqY