EgoMon Gaze and Video Dataset for Visual Saliency Prediction