Object Detection on T_OD
74.1mIoUSFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SFTBackbone=Qwen2-VL, Prompting Strategy=Multi-turn Prompt2026.04 | 74.1 | 6.5 | |
| BaseBackbone=Qwen2-VL, Prompting Strategy=Multi-turn Prompt2026.04 | 73.9 | 20.4 | |
| SKRBackbone=Qwen2-VL, Prompting Strategy=Multi-turn Prompt2026.04 | 72.6 | 4.8 | |
| BaseBackbone=Qwen2-VL, Prompting Strategy=One-turn Prompt2026.04 | 62.8 | 6.8 | |
| SKRBackbone=LLaVA-1.6, Localization Strategy=OCR2026.04 | 61.4 | 1.6 | |
| BaseBackbone=LLaVA-1.6, Localization Strategy=OCR2026.04 | 59.9 | 10.3 | |
| SKRBackbone=Phi-3-VL, Localization Strategy=OCR2026.04 | 34.4 | 0.9 | |
| BaseBackbone=Phi-3-VL, Localization Strategy=OCR2026.04 | 33.6 | 30.7 | |
| SKRBackbone=LLaVA-1.6, Localization Strategy=Fuzzy Location2026.04 | 26.6 | 1.7 | |
| SKRBackbone=Phi-3-VL, Localization Strategy=Fuzzy Location2026.04 | 23.2 | 0.8 | |
| BaseBackbone=Phi-3-VL, Localization Strategy=Fuzzy Location2026.04 | 16.4 | 35 | |
| BaseBackbone=LLaVA-1.6, Localization Strategy=Fuzzy Location2026.04 | 14.2 | 41 | |
| SFTBackbone=LLaVA-1.6, Localization Strategy=Fuzzy Location2026.04 | 6.5 | 6.4 | |
| SFTBackbone=Phi-3-VL, Localization Strategy=Fuzzy Location2026.04 | 5.6 | 7.1 |