In this example, the user specifies the I-AUV an Object Recovery Intervention. The loaded image is a photo-mosaic obtained by a real AUV (from the Systems, Robotics and Vision group, University of the Balearic Islands). The user wants to recover the handsaw, and uses some GUI's functions to detect it. The common segmentation algorithm can not detect the target, so the user crops the image to make easier the target detection. Then the user sets the best fitting ellipse and the grasping points. All the data is exported in to a XML file by the user. This file will be converted by a compiler into a specific language for the robot.