Infrastructure improvement proposal on BDD100K (test)
69.17RecallOD-RASE (Long-CLIP)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| OD-RASE (Long-CLIP)Vision Encoder=Long-CLIP, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 69.17 | 84.31 | 76 | 44.76 | |
| OD-RASE (CLIP)Vision Encoder=CLIP, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 66.86 | 84.87 | 74.8 | 43.17 | |
| OD-RASE (ViT-B)Vision Encoder=ViT-B, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 62.21 | 86.12 | 72.24 | 40.62 | |
| OD-RASE (ResNet-50)Vision Encoder=ResNet-50, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 59.44 | 79.64 | 68.08 | 33.2 | |
| Phi-3Zero-shot=true2026.03 | 51.35 | 21.6 | 28.73 | 18.24 | |
| GPT-4oZero-shot=true2026.03 | 50.73 | 21.54 | 29.38 | 20.4 | |
| LLaVA-1.5Zero-shot=true2026.03 | 46.71 | 26.91 | 27.47 | 19.62 | |
| Qwen2-VLZero-shot=true2026.03 | 27.88 | 14.47 | 18.49 | 11.65 | |
| InternVL2Zero-shot=true2026.03 | 25.11 | 12.21 | 15.62 | 9.79 |