eel4759_classification - Comparison of different ImageNet-based CNN models for classifying diabetic retinopathy images

Age	Commit message (Collapse)	Author	Files	Lines
2026-04-21	main: increase batch size	Vineet Kumar	1	-1/+1

2026-04-20	increase resolution to 384px and soften oversampler weights	Vineet Kumar	1	-1/+3
	- Default img_size changed from 224 to 384 (already plumbed through build_transforms/build_dataloaders, just needed passing from main.py) - WeightedRandomSampler now uses sqrt(1/count) instead of 1/count to reduce over-prediction of minority classes against visually similar majority class (Healthy vs Mild NPDR) - Default batch size reduced 64->32 to accommodate larger images - 4-epoch warmup-only test: macro F1 0.45, weighted F1 0.74, kappa 0.62 already exceeding previous 22-epoch best on all metrics
2026-04-20	improve training for class-imbalanced DR classification	Vineet Kumar	1	-17/+43
	- WeightedRandomSampler for balanced class sampling during training - FocalLoss (gamma=2) for hard-example mining, no alpha (sampler handles balance) - Stronger augmentation: RandomResizedCrop, ColorJitter, GaussianBlur, 30deg rotation - Discriminative LR: backbone at 0.1x, head at 1x - 3-epoch linear warmup + cosine decay scheduler - Early stopping on macro F1 instead of weighted F1 - Updated defaults: batch_size=64, epochs=50, patience=10 - Comparison chart now shows both macro and weighted F1 - DataParallel checkpoint handling (strip module. prefix) - Enable cudnn.benchmark for training speed
2026-04-18	initial commit	Vineet Kumar	1	-0/+161