no code implementations • 22 Jan 2022 • Maliha Arif, Calvin Yong, Abhijit Mahalanobis
Even though convolutional neural networks can classify objects in images very accurately, it is well known that the attention of the network may not always be on the semantically important regions of the scene.
