AnemiaVision fine-tunes EfficientNet-B3 with augmentations like TrivialAugmentWide and Mixup to achieve 96.2% validation accuracy and 0.98 AUC-ROC for non-invasive anemia detection from smartphone images.
Random erasing data augmentation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
AnemiaVision: Non-Invasive Anemia Detection via Smartphone Imagery Using EfficientNet-B3 with TrivialAugmentWide, Mixup Augmentation, and Persistent Patient History Management
AnemiaVision fine-tunes EfficientNet-B3 with augmentations like TrivialAugmentWide and Mixup to achieve 96.2% validation accuracy and 0.98 AUC-ROC for non-invasive anemia detection from smartphone images.