Infrastructure improvement proposal on BDD100K (val)
70.22RecallOD-RASE (Long-CLIP)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| OD-RASE (Long-CLIP)Vision Encoder=Long-CLIP, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 70.22 | 83.68 | 76.36 | 45.79 | |
| OD-RASE (CLIP)Vision Encoder=CLIP, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 67.66 | 84.51 | 75.15 | 43.63 | |
| OD-RASE (ViT-B)Vision Encoder=ViT-B, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 63 | 85.46 | 72.53 | 41 | |
| OD-RASE (ResNet-50)Vision Encoder=ResNet-50, Text Encoder=RoBERTa-Base, Zero-shot=true2026.03 | 60.25 | 79.45 | 68.53 | 34.57 | |
| GPT-4oZero-shot=true2026.03 | 54.26 | 22.79 | 31.19 | 21.64 | |
| Phi-3Zero-shot=true2026.03 | 52.15 | 21.69 | 28.94 | 18.38 | |
| LLaVA-1.5Zero-shot=true2026.03 | 47.21 | 26.39 | 27.27 | 19.5 | |
| Qwen2-VLZero-shot=true2026.03 | 28 | 14.34 | 18.4 | 11.64 | |
| InternVL2Zero-shot=true2026.03 | 25.69 | 12.49 | 15.91 | 10.01 |