A method for tracking an object that is embedded within images of a scene, including: in a sensor unit, generating, storing and transmitting over a communication link a succession of images of a scene. In a remote control unit, receiving the images, receiving a command for selecting an object of interest in a given image and determining object data associated with the object and transmitting the object data to the sensor unit. In the sensor unit, identifying the given image and the object of interest using the object data, and tracking the object in other images. If the object cannot be located in the latest image of the stored succession of images, using information of images in which the object was located to predict estimated real-time location thereof and generating direction commands to the movable sensor for generating realtime images of the scene and locking on the object.