Zero-Shot Transfer Image Classification 벤치마크
Zero-Shot Transfer Image Classification on Food-101
Top 1 Accuracy
- 2023-03-23 — MAWS (ViT-2B): Top 1 Accuracy 96.2
| Rank | Model | Top 1 Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | MAWS (ViT-2B) | 96.2 | The effectiveness of MAE pre-pretraining for billion-scale pretraining | facebookresearch/maws | 2023 |
| 2 | EVA-CLIP-18B | 95.8 | EVA-CLIP-18B: Scaling CLIP to 18 Billion Parameters | baaivision/EVA · baaivision/eva | 2024 |
| 3 | InternVL-C | 95.3 | InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks | opengvlab/internvl · opengvlab/internvl-mmdetseg | 2023 |
| 4 | EVA-CLIP-E/14+ | 94.9 | EVA-CLIP: Improved Training Techniques for CLIP at Scale | baaivision/eva · PaddlePaddle/PaddleMIX · Yui010206/CREMA · +1 | 2023 |
| 5 | Diffusion Classifier (zero-shot) | 77.7 | Your Diffusion Model is Secretly a Zero-Shot Classifier | diffusion-classifier/diffusion-classifier · SamsungSAILMontreal/ForestDiffusion · tajamul21/fate · +1 | 2023 |