Add bounding box to the detected object

[GCXLLY]

A cool project! It perfectly addresses the challenge of running YOLO on Hololens, which OpenCVforUnity struggles with. While exploring your project, I was curious about the object detection feature. Specifically, is there a method within your framework that allows for the visualization of detected objects through bounding boxes instead of using that blue tag shown in the demo? Really appreciate the hard work you've put into this project and looking forward to your reply.

[LocalJoost]

Hi @GCXLLY, Thanks for your compliments. But what you need the understand is how the project works:

• The HoloLens takes a 2D picture of the view
• I let YOLO find the object. This gives a bounding box all right - a 2D one, on the picture. Not in space
• I pretend the picture is hanging before, about a meter.
• I shoot a ray from the camera through the center of the box. If the ray hits something, it is most likely the object.

So I have no idea of the dimensions of the object. A bounding box is most likely very off. You might try to use something else as a labeler, but I doubt it will be very accurate. And it will 2D, because I have no data about the actual 3D dimensions of the object

[GCXLLY]

That makes a lot of sense. Thanks for your clear explanation!

[LocalJoost]

You are welcome!