Get (x, y, z) world coordinates from a point cloud from an RGB Image
Hello,
I got a ZED 2 stereo camera and I am subscribing to the left camera's RGB Image and to the generated point cloud. With the RGB Image, I can perform object detection and get a set of (X,Y) coordinates describing the center of objects. Out of those (X,Y) image coordinates, how can I obtain the (X,Y,Z) world coordinates of the detected objects out of the registered point cloud? E.g.: How can I obtain the world coordinates of the point with (X,Y) image coordinates? I am using ROS Melodic, if it's relevant.
Asked by Pedr on 2021-09-20 14:02:31 UTC
Comments
The transformation from 2D to 3D is not trivial, you need to know the position and orientation of your camera(s) for example in the World Coordinates. There are packages that does heavy lifting for you with object detection (6D pose - position and orientation). Take a look at this: https://github.com/NVlabs/Deep_Object_Pose
Asked by osilva on 2021-09-20 16:47:42 UTC
I think this is a duplicate of #q372663, and #q367829, If you think that it is not duplicate then write, will try my best to answer your question. With this answer, you can also write the same code for ZED 2 camera also.
If you think your question is duplicate then close it.
Thanks
Asked by Ranjit Kathiriya on 2021-09-20 17:27:40 UTC
if you have any other questions regarding this topic, feel free to drop a comment over there.
Asked by Ranjit Kathiriya on 2021-09-21 05:14:00 UTC