Classification on FGVC Aircraft
57Top-1 AccuracyPEcore
Evaluation Results
| Method | Links | |
|---|---|---|
| PEcoreData size=5B, Zero-shot=true2026.03 | 57 | |
| SigLIP2Data size=10B, Zero-shot=true2026.03 | 54.8 | |
| SigLIPData size=10B, Zero-shot=true2026.03 | 44 | |
| LlipData size=2.5B, Zero-shot=true2026.03 | 41.5 | |
| ITOPre-training Dataset=DataComp-1B, Backbone=ViT-B/16, Training Epochs=10 epochs, Zero-shot protocol=EVA-CLIP2026.03 | 30.8 | |
| MetaCLIPData size=2.5B, Zero-shot=true2026.03 | 30.3 | |
| ITO sub2Pre-training Dataset=DataComp-1B, Backbone=ViT-B/16, Training Epochs=10 epochs, Zero-shot protocol=EVA-CLIP2026.03 | 27.2 | |
| LSA+OnZetaBackbone=ViT-B/162026.06 | 26.67 | |
| LSABackbone=ViT-B/162026.06 | 26.28 | |
| OpenCLIPData size=2B, Zero-shot=true2026.03 | 25.8 | |
| CLIPPre-training Dataset=DataComp-1B, Backbone=ViT-B/16, Training Epochs=10 epochs, Zero-shot protocol=EVA-CLIP2026.03 | 25.4 | |
| OnZetaBackbone=ViT-B/162026.06 | 24.81 | |
| TPTBackbone=ViT-B/162026.06 | 24.78 | |
| CLIP baselineBackbone=ViT-B/162026.06 | 24.36 | |
| LSA+OnZetaBackbone=ResNet-502026.06 | 18.88 | |
| LSABackbone=ResNet-502026.06 | 18.12 | |
| TPTBackbone=ResNet-502026.06 | 17.58 | |
| OnZetaBackbone=ResNet-502026.06 | 17.26 | |
| CLIP baselineBackbone=ResNet-502026.06 | 16.92 | |
| DreamLIPData size=30M, Zero-shot=true2026.03 | 7.9 | |
| FLAIRData size=30M, Zero-shot=true2026.03 | 7.2 | |
| GoldiCLIPData size=30M, Zero-shot=true2026.03 | 6.6 | |
| COSMOSData size=30M, Zero-shot=true2026.03 | 5.9 | |
| CLIPData size=30M, Zero-shot=true2026.03 | 5.7 | |
| SigLIPData size=30M, Zero-shot=true2026.03 | 5.7 | |
| ITO sub2Backbone=ViT-B/16, Pre-training Dataset=CC3M-recap, Evaluation Protocol=Zero-shot2026.03 | 2.6 | |
| FLAIRBackbone=ViT-B/16, Pre-training Dataset=CC3M-recap, Evaluation Protocol=Zero-shot2026.03 | 1.6 | |
| ITO sub3Backbone=ViT-B/16, Pre-training Dataset=CC3M-recap, Evaluation Protocol=Zero-shot2026.03 | 1.2 |