In this paper we introduce adaptive vision techniques used, for example, in video-conferencing applications. Radial Basis Function (RBF) networks have been trained for gesture-based communication with colour/motion cues to direct face detection and capture ?attentional frames ?. These focus the processing for Visually Mediated Interaction via an ppearance-based approach with Gabor filter coefficients use as input to time-delay RBF networks. We use these methods for behaviour (user-camera) coordination in an integrated system.