YoLoWithHoloLens

[Xyy1209]

Hello. I saw the video you put on the Youtube named "Object Detection With YOLOv2". It's realy amazing project! I am doing something similar to you. I used the WindowsMachineLearning platform to run TinyYolo(only detect 20 categories) On HoloLens . however, there is some problem occured in the progress. I want to map the bounding box detected by YOLO on the image to the 3D real sense, but I don't know how to map the pixel on the image to 3D world coordinate. Could you give me some advice or any relative code ? Thank you very much!

[tarukosu]

Thank you for your comment.
I convert the 2D position (center of bounding box) in image to a Raycast in 3D.
3D position of the detected object is the position where the Raycast hits spatial mapping.
This blog is helpful, though it's not mine.
https://qiita.com/rakusan/items/6af7d35dc7b0222952fa