Image Classification on Food101 (Clean Accuracy)
87.2Clean AccuracyRAR
Evaluation Results
| Method | Links | |
|---|---|---|
| RARBase Model=Qwen-VL2024.03 | 87.2 | |
| RARBase Model=LLaVA1.52024.03 | 86.2 | |
| RARBase Model=Intern-IXC22024.03 | 83.9 | |
| CLIPYear=20212026.04 | 83.84 | |
| CLIPBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.03 | 83.61 | |
| GPT-4V2024.03 | 80.1 | |
| AGFTBackbone=ViT-B/32, Evaluation Protocol=Zero-shot, Training Strategy=Adversarially fine-tuned2026.03 | 72.14 | |
| OursModel Configuration=1 tree2026.04 | 70.31 | |
| PMG-AFTBackbone=ViT-B/32, Evaluation Protocol=Zero-shot, Training Strategy=Adversarially fine-tuned2026.03 | 70.03 | |
| OursModel Configuration=5 trees2026.04 | 69.76 | |
| AoSYear=20252026.04 | 69.53 | |
| GLADIATORBackbone=ViT-B/32, Evaluation Protocol=Zero-shot, Training Strategy=Adversarially fine-tuned2026.03 | 67.32 | |
| FAREYear=20242026.04 | 67 | |
| PMG-FTYear=20242026.04 | 61.26 | |
| TeCoABackbone=ViT-B/32, Evaluation Protocol=Zero-shot, Training Strategy=Adversarially fine-tuned2026.03 | 56.24 | |
| TGA-ZSRBackbone=ViT-B/32, Evaluation Protocol=Zero-shot, Training Strategy=Adversarially fine-tuned2026.03 | 51.31 | |
| TeCoAYear=20232026.04 | 47.27 |