A hard attention model pretrained on logit-ranked patches reaches 75% top-1 and 91% top-5 on ImageNet while classifying from under one-third of the image.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
Saccader: Improving Accuracy of Hard Attention Models for Vision
A hard attention model pretrained on logit-ranked patches reaches 75% top-1 and 91% top-5 on ImageNet while classifying from under one-third of the image.