Object Classification on Visual Genome Objects top-150 vocabulary
29.7Top-1 AccuracyScenarioCLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ScenarioCLIPKnowledge Distillation=true, Evaluation=Zero-shot2025.11 | 29.7 | 50.13 | 57.7 | |
| CLIPBackbone=ViT-B/32, Evaluation=Zero-shot2025.11 | 29.43 | 55.98 | 66.27 | |
| ScenarioCLIP w/o KDKnowledge Distillation=false, Evaluation=Zero-shot2025.11 | 29.13 | 49.16 | 56.69 | |
| PyramidCLIPEvaluation=Zero-shot2025.11 | 21.94 | 45.52 | 56.17 |