Fetching the paper…

Deep Gaze I: Boosting Saliency Prediction with Feature Maps Trained on ImageNet · Around