Adding three small convolutional attention blocks to ResNet-18 improves classification accuracy by up to 1.5 percentage points on three benchmark image datasets.
Dual attention network for scene segmentation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Vision Eagle Attention: a new lens for advancing image classification
Adding three small convolutional attention blocks to ResNet-18 improves classification accuracy by up to 1.5 percentage points on three benchmark image datasets.