A device for acquiring a combined eye gaze image of an object under dark-eye effect conditions, with a first camera, a second camera, a first light source and a second light source being located on opposite sides of, and on essentially equal distance to a central optical axis. A control unit is arranged to acquire the combined eye gaze image by capturing, in a first point in time, a first frame of the object with the first camera with the second light source activated, and, at a second point in time, capturing a second frame of the object with the second camera with the first light source activated. The device comprises no additional light sources further away from the central optical axis than the first location and the second location. The device is thereby designed as compact as possible while a sufficient eye gaze tracking accuracy and robustness is maintained.
