Domain Generalization
The idea of Domain Generalization is to learn from one or multiple training domains, to extract a domain-agnostic model which can be applied to an unseen domain
Source: Diagram Image Retrieval using Sketch-Based Deep Learning and Transfer Learning
Papers
Showing 1–10 of 1751 papers
All datasetsPACSVizWiz-ClassificationImageNet-COffice-HomeImageNet-AImageNet-RDomainNetVLCSTerraIncognitaGTA-to-Avg(Cityscapes,BDD,Mapillary)ImageNet-SketchGTA5 to Cityscapes
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | Model soups (BASIC-L) | Top-1 accuracy | 77.18 | — | Unverified |
| 2 | Model soups (ViT-G/14) | Top-1 accuracy | 74.24 | — | Unverified |
| 3 | CAR-FT (CLIP, ViT-L/14@336px) | Top-1 accuracy | 65.5 | — | Unverified |
| 4 | ConvNeXt-XL (Im21k, 384) | Top-1 accuracy | 55 | — | Unverified |
| 5 | CAFormer-B36 (IN21K, 384) | Top-1 accuracy | 54.5 | — | Unverified |
| 6 | LLE (ViT-H/14, MAE, Edge Aug) | Top-1 accuracy | 53.39 | — | Unverified |
| 7 | ConvFormer-B36 (IN21K, 384) | Top-1 accuracy | 52.9 | — | Unverified |
| 8 | CAFormer-B36 (IN21K) | Top-1 accuracy | 52.8 | — | Unverified |
| 9 | ConvFormer-B36 (IN21K) | Top-1 accuracy | 52.7 | — | Unverified |
| 10 | MAE (ViT-H, 448) | Top-1 accuracy | 50.9 | — | Unverified |