List your company
ResearchFormerly WizWand

Image Classification on Spawrious O2O-Hard

95.7AccuracyDFR
35.951.42566.9582.475May 6, 2026May 10, 2026May 15, 2026May 19, 2026May 24, 2026May 28, 2026Jun 2, 2026
Updated 5mo ago

Evaluation results

MethodLinks
DFRBackbone=ViT-G (1.1B), Note=Requires group labels2026.0695.7
DFR + A2Backbone=ViT-S (21M), Note=Requires group labels2026.0695.2
DFRBackbone=ViT-L (300M), Note=Requires group labels2026.0694.3
Attending on Attention (A2)Backbone=ViT-S (21M), Feature Extractor=DINOv22026.0691
DFRBackbone=ViT-B (86M), Note=Requires group labels2026.0690.1
OpenCLIP ZS prompt (full)Backbone=ViT-L-14, Feature Extractor=OpenCLIP, Evaluation=Zero-shot prompt2026.0689.8
A2 cross-modelBackbone=ViT-S x ViT-L, Feature Extractor=DINOv22026.0689.7
Attending on Attention (A2)Backbone=ViT-B (86M), Feature Extractor=DINOv22026.0688.4
Full onlyBackbone=ViT-G (1.1B), Feature Extractor=DINOv22026.0687.8
Full only (518^2 image)Backbone=ViT-L (300M), Resolution=518x518, Feature Extractor=DINOv22026.0686.3
Full only (518^2 image)Backbone=ViT-S (21M), Resolution=518x518, Feature Extractor=DINOv22026.0686.2
DFRBackbone=ViT-S (21M), Note=Requires group labels2026.0686.1
MMD + AnticlusteringBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=Anticlustering2026.0585.9
MMD + ORDEREDBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=ORDERED2026.0585.1
Attending on Attention (A2)Backbone=ViT-G (1.1B), Feature Extractor=DINOv22026.0683.1
Full only (518^2 image)Backbone=ViT-B (86M), Resolution=518x518, Feature Extractor=DINOv22026.0682.5
Scaling on ScalesBackbone=ViT-S (21M), Feature Extractor=DINOv2, scales=[1, 2]2026.0682.2
Attending on Attention (A2)Backbone=ViT-L (300M), Feature Extractor=DINOv22026.0682.1
Full onlyBackbone=ViT-L (300M), Feature Extractor=DINOv22026.0680.8
Full onlyBackbone=ViT-S (21M), Feature Extractor=DINOv22026.0680.5
CausIRLBackbone=ResNet-50, Training Dataset=Spawrious2026.0680.4
Center CropBackbone=ViT-S (21M), Feature Extractor=DINOv22026.0680.2
CORALBackbone=ResNet-50, Training Dataset=Spawrious2026.0679.65
CORAL + VaRDASSBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=VaRDASS2026.0579.6
Full onlyBackbone=ViT-B (86M), Feature Extractor=DINOv22026.0678.8
CORAL + ORDEREDBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=ORDERED2026.0578.1
GroupDROBackbone=ResNet-50, Training Dataset=Spawrious, Note=Requires group labels2026.0676.99
IRMBackbone=ResNet-50, Training Dataset=Spawrious, Note=Requires group labels2026.0674.9
CDAN + SDATBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0573.7
CORAL + AnticlusteringBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=Anticlustering2026.0573.7
MMD + VaRDASSBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=VaRDASS2026.0572.7
Center CropBackbone=ViT-G (1.1B), Feature Extractor=DINOv22026.0672.6
ERMBackbone=ResNet-50, Training Dataset=Spawrious2026.0671.32
DANNBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0571.1
Full onlyBackbone=ViT-L-14, Feature Extractor=OpenCLIP2026.0671
CDANBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0570.3
MMD + k-means++Backbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=k-means++2026.0568.4
CORAL + DPPBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=DPP2026.0567.8
CDAN + ELSBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0567.3
MMDBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0565.5
CORALBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0564.1
MMD + DPPBackbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=DPP2026.0563.5
ERMBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0562.1
CORAL + k-means++Backbone=ResNet-18, Optimizer=Adam, Iterations=3000, Sampler=k-means++2026.0561.4
ARMBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0560.6
MCCBackbone=ResNet-18, Optimizer=Adam, Iterations=30002026.0554
Random CropsBackbone=ViT-S (21M), Feature Extractor=DINOv22026.0638.2