Saccader: Improving Accuracy of Hard Attention Models for Vision

Open in new window