Medical Image Classification on OCTMNIST
91.7AccuracyMO-MAE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MO-MAE2026.05 | 91.7 | 0.993 | |
| MoCoV3Backbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 79.96 | — | |
| SaEBudget=20%, Number of seeds=52026.02 | 79.8 | — | |
| NNCLRBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 79.16 | — | |
| ReSSLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 78.56 | — | |
| MedVIT-S2026.05 | 78.2 | 0.96 | |
| ResNet-50Resolution=2242026.02 | 77.6 | 0.958 | |
| ResNet-502026.05 | 77.6 | 0.958 | |
| Barlow TwinsBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 77.28 | — | |
| Google AutoML Vision2026.02 | 77.1 | 0.963 | |
| Google AutoML2026.05 | 77.1 | 0.963 | |
| MedVIT-T2026.05 | 76.7 | 0.961 | |
| ResNet-18Resolution=2242026.02 | 76.3 | 0.958 | |
| AutoKeras2026.02 | 76.3 | 0.955 | |
| ResNet-182026.05 | 76.3 | 0.958 | |
| AutoKeras2026.05 | 76.3 | 0.955 | |
| ResNet-50Resolution=282026.02 | 76.2 | 0.952 | |
| MedCoOp + EntropyBudget=20%, Number of seeds=52026.02 | 76.2 | — | |
| MedVIT-L2026.05 | 76.1 | 0.945 | |
| MedCoOp + BADGEBudget=20%, Number of seeds=52026.02 | 75.3 | — | |
| VICRegBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 75.2 | — | |
| BYOLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 74.94 | — | |
| ResNet-18Resolution=282026.02 | 74.3 | 0.943 | |
| AMIMV-SSLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 72 | — | |
| SimCLRBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 68.82 | — | |
| MedCoOp + CoresetBudget=20%, Number of seeds=52026.02 | 67.4 | — | |
| BiomedCoOpBudget=20%, Number of seeds=5, inference_mode=few-shot reference2026.02 | 66.93 | — | |
| PCBBudget=20%, Number of seeds=52026.02 | 66.8 | — | |
| RandomBudget=20%, Number of seeds=52026.02 | 65.2 | — | |
| BioVLMTrain Params=30K2026.04 | 62.57 | — | |
| MaPLeTrain Params=5.3M2026.04 | 60.76 | — | |
| DINOBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 60.7 | — | |
| CoCoOpTrain Params=44K2026.04 | 60.47 | — | |
| auto-sklearn2026.02 | 60.1 | 0.887 | |
| Auto-sklearn2026.05 | 60.1 | 0.887 | |
| TCPTrain Params=0.5M2026.04 | 59.92 | — | |
| PromptSRCTrain Params=69K2026.04 | 59.35 | — | |
| CoOpTrain Params=3K2026.04 | 59.2 | — | |
| KgCoOpTrain Params=3K2026.04 | 57.67 | — | |
| G-LoG bi-filtration (MLP)Sigma (σ)=0.52026.02 | 56.7 | 0.872 | |
| G-LoG bi-filtration (MLP)Sigma (σ)=12026.02 | 54.8 | 0.869 | |
| BiomedCoOpTrain Params=3K2026.04 | 54.77 | — | |
| G-LoG bi-filtration (MLP)Sigma (σ)=1.52026.02 | 52.9 | 0.854 | |
| G-LoG bi-filtration (MLP)Sigma (σ)=02026.02 | 52.2 | 0.859 | |
| ProGradTrain Params=3K2026.04 | 47.03 | — | |
| Topo-Med (MLP)2026.02 | 45 | 0.71 | |
| BioMedCLIPTrain Params=02026.04 | 36.8 | — | |
| AMIMV-SSLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9659 | |
| Barlow TwinsBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9705 | |
| BYOLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9567 | |
| DINOBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9001 | |
| MoCoV3Backbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9681 | |
| NNCLRBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9711 | |
| ReSSLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9607 | |
| SimCLRBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.9217 | |
| VICRegBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 0.966 |