Hi folks,

I have an OAK-D Pro camera and I need to detect the pose of multiple items simultaneously instead of just one item then I think a DOPE approach based on ROS may be suited to my requirements. Because such a system uses an RGB image to recognize the object and then calculates its 6DOF pose, I wonder if depth-ai ROS already support it? And do you guys have a guide for it?

Thanks!

Hi @astroquavo
I doubt we have any examples since achieving 6DOF hasn't been done yet as we mostly only use and train models that are capable of standard object detection (x, y, w, h) and not the rotation detection.

cc @Luxonis-Adam

Hi, as of this moment there is no example for 3D pose estimation, we only can provide spatial bounding boxes for detections.