Hi erik
I may not have explained the problem I encountered clearly. My goal is to rotate the camera clockwise by 90 degrees and install it to detect the position data (xyz) of the face and the posture angle data (RPY) of the head.
For RGB images and depth maps, I rotated them separately using ImageManip. However, after visualizing bbox, it was found that the ROI area was not on the face.
here is the mre
Thanks in advance!

    Hi @walawala111
    I think this might be an issue of imageManip not passing the rotation information to the spatialLocationCalculator.

    Would have to check that with the team.

    Thanks,
    Jaka

      6 days later

      @walawala111 200 lines of code is not an MRE. Please distil it down to minimal repro example and we can take a look into it.

        Hi walawala111
        It's manip issue. In order to properly set the spatial bounding box of a rotated (transformed) image, you would need the order of operations that were applied to the start image, the undistortion mesh; as well as intrinsics of both stereo as well as color cameras. Current implementation of ImageManip node apparently does not allow this information passthrough.
        DepthaiV3 aims to fix this issue.

        Thanks,
        Jaka