Revision history [back]

The frame of the pose openCV detects is already the camera frame, so you will just need to convert the tvec and rvec into a position and orientation repsectively.

The tvec is simply a translation vector so it's elements can be copied directly to the x, y & z values of the position in a pose message. Rvec is slightly more complicated however, you will need to use the openCV Rodrigues function to convert this into a 3x3 roation matrix. Then put the 9 values from the cv::Mat object into an equivalent tf2:::Matrix3x3 object. This object has a method getRotation which will return a quaternion of this rotation, now you have this quaternion you can copy its elements into the pose method.

Hope this makes sense.