Domain Generalization On Office Home

评估指标

Average Accuracy

评测结果

各个模型在此基准测试上的表现结果

		Paper Title	Repository
MoA (OpenCLIP, ViT-B/16)	90.6	Domain Generalization Using Large Pretrained Models with Mixture-of-Adapters
PromptStyler (CLIP, ViT-L/14)	89.1	PromptStyler: Prompt-driven Style Generation for Source-free Domain Generalization
UniDG + CORAL + ConvNeXt-B	88.9	Towards Unified and Effective Domain Generalization
SIMPLE+	87.7	SIMPLE: Specialized Model-Sample Matching for Domain Generalization	-
VL2V-SD (CLIP, ViT-B/16)	87.38	Leveraging Vision-Language Models for Improving Domain Generalization in Image Classification
CAR-FT (CLIP, ViT-B/16)	85.7	Context-Aware Robust Fine-Tuning	-
GMDG (RegNetY-16GF, SWAD)	84.7	Rethinking Multi-domain Generalization with A General Learning Objective
SIMPLE	84.6	SIMPLE: Specialized Model-Sample Matching for Domain Generalization	-
Ensemble of Averages (RegNetY-16GF)	83.9	Ensemble of Averages: Improving Model Selection and Boosting Performance in Domain Generalization
PromptStyler (CLIP, ViT-B/16)	83.6	PromptStyler: Prompt-driven Style Generation for Source-free Domain Generalization
SPG (CLIP, ViT-B/16)	83.6	Soft Prompt Generation for Domain Generalization
MIRO (RegNetY-16GF, SWAD)	83.3	Domain Generalization by Mutual-Information Regularization with Pre-trained Models
D-Triplet(RegNetY-16GF)	82.6	Domain-aware Triplet loss in Domain Generalization
GMDG (RegNetY-16GF)	80.8	Rethinking Multi-domain Generalization with A General Learning Objective
SEDGE+	80.7	Domain Generalization using Pretrained Models without Fine-tuning	-
Ensemble of Averages (ResNeXt-50 32x4d)	80.2	Ensemble of Averages: Improving Model Selection and Boosting Performance in Domain Generalization
CADG	79.9	CADG: A Model Based on Cross Attention for Domain Generalization	-
SEDGE	79.9	Domain Generalization using Pretrained Models without Fine-tuning	-
GMoE-S/16	74.2	Sparse Mixture-of-Experts are Domain Generalizable Learners
SPG (CLIP, ResNet-50)	73.8	Soft Prompt Generation for Domain Generalization

0 of 45 row(s) selected.