Towards Ignoring Backgrounds and Improving Generalization: a Costless DNN Visual Attention Mechanism

Open in new window