Spatial and temporal visual attention prediction in videos using eye movement data