Nuitrack | Video tutorial #2: Object Detection with Nuitrack AI

Where do I check the library for what object it can detect or recognize? and how do I add other object to detect in the library? such as I want the camera to detect a ball or a ping pong paddle.

Thank you

Hello @hizral .
Now available:

The following types of objects can be detected on a color stream:

  • bottle
  • cigarette
  • phone
  • laptop
  • bag
  • book


and how do I add other object to detect in the library?

At the moment, adding your own objects is not provided, but it is possible that such an opportunity will be added in the future. How would you like to add your object, do you have one picture or many pictures with such balls?

What I’m trying to do is, I would like to make my camera detect a ping pong ball being detect while being played by user, and from the tracking convert it into a 3d point so it can be detected in my game engine on where the location and position of the ball in 3d world.

and can the camera detect object base on color? for example if I have multiple color of balls, can I assign it to detect a specific color in the ball pool?

No. The camera cannot detect objects by color

Hello @Stepan.Reuk
I want to recognize the ball (Intel D435). It’s possible? What other object types are supported?

Hello @Sergiy
No, at the moment. But this is in our plans

Hi @Sergiy, thank you for your question.
Thinking on possible extensions of Nuitrack object detection module:
are you interested in ball detection with a specific/particular appearance such as basketball ball or kind of generic detection of any ball / spherical objects ?

Hi, @TAG
I want to do something like this

So I need to detect any circular object.
Of course, I can dictate some requirements of what this object should be so that the system reliably detects it.

PS I have experience to creating interactive games with laser beam recognition.
But I used a regular web camera with filter.
Not D435 and SDK RealSense / Nuitrack.

I think it will be enough for me to find the fact / event that the coordinates of the round object match with the surface of the screen (floor).