On the Simultaneous Interpretation of Real World Image Sequences and their Natural Language Description.

On the Simultaneous Interpretation of Real World Image Sequences and their Natural Language Description.: "The aim of previous attempts at connecting vision systems and natural language systems has been to provide a retrospective description of the analysed image sequence. The step from such an a posteriori approach towards simultaneous natural language description reveals a problem which has not yet been dealt with in generation systems. Automatic generation of simultaneous descriptions calls for the application of an incremental event recognition strategy and for the adequate coordination of event recognition and language production. In order to enable free interaction between these processes, it is useful to implement them in parallel. In this paper the system SOCCER will be presented, which is based upon such a conception. Short sections of soccer games have been chosen as the domain of discourse. In analogy to radio reports, the system generates a description of the game which it is watching and which the listener cannot see. "