Few Shot Image Classification On Bongard Hoi
평가 지표
Avg. Accuracy
평가 결과
이 벤치마크에서 각 모델의 성능 결과
모델 이름 | Avg. Accuracy | Paper Title | Repository |
|---|---|---|---|
| SVM-Mimic (frozen CLIP RN-50) | 72.45 | Support-Set Context Matters for Bongard Problems | |
| ANIL (ImageNet_R50) | 49.74 | Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions | |
| Meta-Baseline (Scratch_R50) | 54.23 | Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions | |
| Meta-Baseline (ImagNet_R50) | 55.82 | Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions | |
| Meta-Baseline (MoCov2_R50) | 54.30 | Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions | |
| Human (Amateur) | 91.42 | Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions | |
| SVM-Mimic + PMF (fine-tuned CLIP RN-50) | 76.41 | Support-Set Context Matters for Bongard Problems |
0 of 7 row(s) selected.