M. Kümmerer, L. Theis, and M. Bethge. Deep Gaze I: Boosting Saliency Prediction with Feature Maps Trained on ImageNet. ICLR Workshop, June 2015.