Allow image coordinates in CameraDetectionData point-list

[RIFJo]

Describe the feature

According to the documentation, osi3::FeatureData is supposed to contain feature information based on a single sensor. I would like to be able to specify points for a osi3::CameraDetection in image coordinates (x,y) instead of spherical coordinates (azimuth/elevation/distance). Reasoning: The conversion from image coordinates to azimuth and elevation can be arbitrarily complex depending on camera type and lens geometry (which might be unknown to a pure image-based detector), and distance data is not available from a single camera sensor anyway.

Describe the solution you would like

• add a new Vector2D field "imageCoordinates" to osi3:CameraPoint, to allow specification of a 2D point in the image that the camera has produced. A module can fill these fields if the conversion to spherical coordinates is unavailable. (Or two fields "imageX, imageY" with type integer or double)

Describe alternatives you have considered

• relative coordinates (0..1) with respect to image width/height
• use of an alternative list with the 2D coordinates in CameraDetectionData (but allow only one of the lists to be filled to avoid confusion)

Describe the backwards compatibility

Since these are an optional new data fields, no backwards compatibility issues are introduced by this suggestion.

Additional context

I would like to be able to send the results of an image-based object detector as an OSI message. Conversion to azimuth/elevation is impossible since the camera geometry is unknown.