Pretrained CNNs with aggressive class-balanced augmentation report 98.9% binary and 84.6% five-class accuracy on APTOS 2019, but five-class superiority over Topo-CNN holds only for accuracy, not macro precision, recall, or AUC.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
Addressing High Class Imbalance in Multi-Class Diabetic Retinopathy Severity Grading with Augmentation and Transfer Learning
Pretrained CNNs with aggressive class-balanced augmentation report 98.9% binary and 84.6% five-class accuracy on APTOS 2019, but five-class superiority over Topo-CNN holds only for accuracy, not macro precision, recall, or AUC.