Custom Computer Vision with Xreal Air 2 Ultra

Development & Help
1

Hello! Hope all is well.

I am considering purchasing a Xreal Air 2 Ultra for research projects involving running custom computer vision models on images / videos from wearable AR. Does the Xreal Air 2 Ultra and NRSDK support:

(1) retrieving images/videos from the on-device cameras,
(2) running custom CV models such as YOLOv8,
(3) can it position a box or some virtual object on top of the detected objects based on depth estimation?
(4) and can hand detection figure out whether a person is grabbing any of the detected objects?

Please advise, thank you!

1 Like
2

Hi, Blueming.

  1. The image texture cannot be available on XREAL Air 2 Ultra;
  2. Of course you can use your trained model, but the CV data is inaccessible.
  3. You can place virtual objects on detected planes, images, or mesh.
  4. The Hand Tracking function only supports detecting the gesture of your hand.
3

But doesn’t the documentation say we can get the Camera Image data?

4

The camera data is RGB cam instead of CV. The Ultra only comes with CV cams.

5

Will the NReal Light no longer be produced?