Image Classification on EuroSAT (val)
99.17AccuracyStelLA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| StelLAModel=ViT Large, Params (%)=0.542025.10 | 99.17 | — | |
| LoRAModel=ViT Large, Params (%)=0.532025.10 | 99.11 | — | |
| PiSSAModel=ViT Large, Params (%)=0.532025.10 | 99.11 | — | |
| DoRAModel=ViT Large, Params (%)=0.552025.10 | 99 | — | |
| LoRAModel=ViT Base, Params (%)=0.722025.10 | 98.98 | — | |
| PiSSAModel=ViT Base, Params (%)=0.722025.10 | 98.93 | — | |
| DoRAModel=ViT Base, Params (%)=0.752025.10 | 98.91 | — | |
| StelLAModel=ViT Base, Params (%)=0.732025.10 | 98.91 | — | |
| OTTEREvaluation Protocol=Linear Probing, Label Distribution Estimation=BBSE+OT2024.04 | 75.9 | — | |
| Linear Probing baselineEvaluation Protocol=Linear Probing, Label Distribution Estimation=None2024.04 | 74.6 | — | |
| BBSE+PMEvaluation Protocol=Linear Probing, Label Distribution Estimation=BBSE+PM2024.04 | 71.6 | — | |
| OTTEREvaluation Protocol=Zero-Shot, Label Distribution Estimation=BBSE+OT2024.04 | 34 | — | |
| Zero-Shot baselineEvaluation Protocol=Zero-Shot, Label Distribution Estimation=None2024.04 | 32.9 | — | |
| BBSE+PMEvaluation Protocol=Zero-Shot, Label Distribution Estimation=BBSE+PM2024.04 | 19.2 | — | |
| Helber et al.Reference=Helber et al. (2019)2019.11 | — | 98.57 | |
| Imagenet (sup.)Pre-training=ImageNet supervised, Backbone=ResNet-18, Evaluation Protocol=Fine-tuning2021.03 | — | 86.44 | |
| InDomainReference=Ours, Fine-tuning=In-domain2019.11 | — | 99.2 | |
| MoCo-v2Pre-training=MoCo-v2, Backbone=ResNet-18, Evaluation Protocol=Fine-tuning2021.03 | — | 83.72 | |
| MoCo-v2+TPPre-training=MoCo-v2 + Temporal Positional embedding, Backbone=ResNet-18, Evaluation Protocol=Fine-tuning2021.03 | — | 89.51 | |
| Random init.Pre-training=Random initialization, Backbone=ResNet-18, Evaluation Protocol=Fine-tuning2021.03 | — | 63.21 | |
| SeCoPre-training=SeCo (Seasonal Contrast), Backbone=ResNet-18, Evaluation Protocol=Fine-tuning2021.03 | — | 93.14 |