Referring Expression Comprehension on RefCOCOg (Accuracy)
84.3AccuracyOurs
Evaluation Results
| Method | Links | |
|---|---|---|
| OursBackbone=LLaVA-1.5-13B2025.03 | 84.3 | |
| OursBackbone=LLaVA-1.5-7B2025.03 | 82.3 | |
| OursBackbone=DeepSeekVL-1.3B2025.03 | 65.2 | |
| GoM (Object Num IDs + Relation labels)MLM=LlamaV-o1-11B2026.03 | 57.6 | |
| GoM (Object Num IDs + Relation labels)MLM=Qwen-2.5-VL-7B2026.03 | 57.5 | |
| GoM (Object Text IDs + Relation labels)MLM=Qwen-2.5-VL-7B2026.03 | 57.4 | |
| GoM (Object Num IDs)MLM=LlamaV-o1-11B2026.03 | 57.3 | |
| GoM (Object Text IDs)MLM=LlamaV-o1-11B2026.03 | 57.2 | |
| GoM (Object Text IDs + Relation labels)MLM=LlamaV-o1-11B2026.03 | 56.9 | |
| GoM (Object Num IDs)MLM=Qwen-2.5-VL-7B2026.03 | 56.8 | |
| GoM (Object Text IDs)MLM=Qwen-2.5-VL-7B2026.03 | 56.5 | |
| GoM (Object Num IDs)MLM=Gemma-3-4B2026.03 | 56.4 | |
| GoM (Object Text IDs)MLM=Gemma-3-4B2026.03 | 56.3 | |
| GoM (Object Text IDs + Relation labels)MLM=Gemma-3-4B2026.03 | 56.3 | |
| GoM (Object Num IDs + Relation labels)MLM=Gemma-3-4B2026.03 | 56.3 | |
| SoMMLM=Qwen-2.5-VL-7B2026.03 | 55.5 | |
| SoMMLM=LlamaV-o1-11B2026.03 | 55.3 | |
| SoMMLM=Gemma-3-4B2026.03 | 54.8 | |
| BaselineBackbone=LLaVA-1.5-13B2025.03 | 5.28 | |
| BaselineBackbone=LLaVA-1.5-7B2025.03 | 2.92 | |
| BaselineBackbone=DeepSeekVL-1.3B2025.03 | 1.5 |