Referring Expression Comprehension on RefCOCO (test-B)
92.5AccuracyInternVL2.5-78B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| InternVL2.5-78BParameters=78B2024.12 | 92.5 | — | — | |
| VILLA_LARGEdetected proposals=false, Model Scale=LARGE2020.06 | 91.62 | — | — | |
| UNINEXT-HType=Specialist2024.09 | 91.5 | — | — | |
| UNINEXT-H2025.12 | 91.5 | — | — | |
| UNINEXT-HBackbone=Huge2024.12 | 91.5 | — | — | |
| UNINEXT-HModel type=Specialist SOTAs (Specialist/Finetuned)2023.06 | 91.46 | — | — | |
| UNINEXT-HModel type=Specialist SOTAs2023.08 | 91.46 | — | — | |
| UNINEXTModel type=Specialist models2023.10 | 91.46 | — | — | |
| UNINEXT-HType=Specialist2023.11 | 91.46 | — | — | |
| UNINEXT-HModel type=Specialist Models2023.12 | 91.46 | — | — | |
| VILLA_BASEdetected proposals=false, Model Scale=BASE2020.06 | 91.38 | — | — | |
| UNITER_LARGEdetected proposals=false, Model Scale=LARGE2020.06 | 91.19 | — | — | |
| Qwen2-VL-72BType=Generalist2024.09 | 90.7 | — | — | |
| Qwen2-VL-72BParameters=72B2024.12 | 90.7 | — | — | |
| UNITER_BASEdetected proposals=false, Model Scale=BASE2020.06 | 90.46 | — | — | |
| VGent2025.12 | 89.8 | — | — | |
| ONE-PEACEType=Specialist2024.09 | 89.3 | — | — | |
| ONE-PEACE2025.12 | 89.3 | — | — | |
| ONE-PEACE2024.12 | 89.3 | — | — | |
| ONE-PEACEModel type=Specialist SOTAs (Specialist/Finetuned)2023.06 | 89.26 | — | — | |
| ONE-PEACEModel type=Specialist SOTAs2023.08 | 89.26 | — | — | |
| ONE-PEACEType=Specialist2023.11 | 89.26 | — | — | |
| CogVLMType=Generalist2024.09 | 89 | — | — | |
| InternVL3.5-38B2025.12 | 89 | — | — | |
| CogVLM-Grounding-17BParameters=17B2024.12 | 89 | — | — | |
| CogVLM-GroundingType=Generalist2023.11 | 88.99 | — | — | |
| CogVLM-17BModel Size=17B2024.07 | 88.99 | — | — | |
| Ferret-v2Type=Generalist2024.09 | 88.9 | — | — | |
| Ferret-v2-13B2025.12 | 88.9 | — | — | |
| Ferret-v2-13BParameters=13B2024.12 | 88.9 | — | — | |
| InternVL3.5-20B-A4B2025.12 | 88.8 | — | — | |
| InternVL3.5-8B2025.12 | 88.7 | — | — | |
| InternVL2-26BType=Generalist2024.09 | 88.4 | — | — | |
| InternVL2-Llama3-76BBackbone=Llama3, Parameters=76B2024.12 | 88.4 | — | — | |
| G-DINO-LModel type=Specialist SOTAs (Specialist/Finetuned)2023.06 | 88.24 | — | — | |
| G-DINO-LModel type=Specialist SOTAs2023.08 | 88.24 | — | — | |
| G-DINO-LModel type=Specialist models2023.10 | 88.24 | — | — | |
| G-DINO-LType=Specialist2023.11 | 88.24 | — | — | |
| G-DINO-LModel type=Specialist Models2023.12 | 88.24 | — | — | |
| G-DINO-LType=Specialist2024.09 | 88.2 | — | — | |
| Grounding-DINO-L2025.12 | 88.2 | — | — | |
| Grounding-DINO-LBackbone=Large2024.12 | 88.2 | — | — | |
| InternVL3-8B2025.12 | 88 | — | — | |
| InternVL3-14B2025.12 | 87.8 | — | — | |
| TextHawk22024.12 | 87.6 | — | — | |
| Qwen2-VL-7BType=Generalist2024.09 | 87.3 | — | — | |
| Qwen2-VL-7B2025.12 | 87.3 | — | — | |
| Qwen2-VL-7BParameters=7B2024.12 | 87.3 | — | — | |
| MM1.52024.12 | 86.7 | — | — | |
| InternVL3-9B2025.12 | 86.6 | — | — | |
| Griffon v22024.07 | 86.5 | — | — | |
| ASMv2-13BParameters=13B2024.02 | 86.24 | — | — | |
| LyricsModel type=Generalist Models, Backbone=Vicuna-13B2023.12 | 86.03 | — | — | |
| InternVL2.5-8BParameters=8B2024.12 | 85.9 | — | — | |
| CoS-7BModel Size=7B, Fine-tuning protocol=LoRA2024.07 | 85.83 | — | — | |
| LION-12BEvaluation Setting=Fine-tuning Setting2023.11 | 85.57 | — | — | |
| Qwen2.5-VL-7B2025.12 | 85.4 | — | — | |
| Qwen-VL-7BModel type=Generalist Models2023.08 | 85.34 | — | — | |
| Qwen-VLType=Generalist2023.11 | 85.34 | — | — | |
| Qwen-VL-7BModel Size=7B2024.07 | 85.34 | — | — | |
| MiniGPT-v2 (7B)Model type=Generalist models2023.10 | 85.33 | — | — | |
| MiniGPTv2Evaluation Setting=Fine-tuning Setting2023.11 | 85.33 | — | — | |
| MiniGPT-v2 (7B)Resolution=448, Visual token count=256, utilizes larger ViT=true2024.03 | 85.33 | — | — | |
| MiniGPTv2Fine-tuning protocol=LoRA2024.07 | 85.33 | — | — | |
| Qwen-VLType=Generalist2024.09 | 85.3 | — | — | |
| Elysium (7B)Resolution=336, Visual token count=1082024.03 | 84.95 | — | — | |
| LION-4BEvaluation Setting=Fine-tuning Setting2023.11 | 84.82 | — | — | |
| MAttNet:subj(+attr+attn)+loc(+dif)+relfeature=res101-frcn2018.01 | 84.57 | — | — | |
| Qwen-VL-7B-ChatModel type=Generalist Models2023.08 | 84.51 | — | — | |
| Qwen-VLModel type=Generalist Models, Backbone=Qwen-7B2023.12 | 84.51 | — | — | |
| Qwen-VL-7BParameters=7B2024.02 | 84.51 | — | — | |
| Ferret-13BEvaluation Setting=Fine-tuning Setting2023.11 | 84.36 | — | — | |
| Ferret-13BType=Generalist2023.11 | 84.36 | — | — | |
| Ferret-13BModel Size=13B2024.07 | 84.36 | — | — | |
| Ferret-13BParameters=13B2024.02 | 84.36 | — | — | |
| MiniGPT-v2 (7B)-chatModel type=Generalist models2023.10 | 84.3 | — | — | |
| MiniGPT-v2-7BParameters=7B2024.02 | 84.3 | — | — | |
| MAttNet:subj(+attr)+loc(+dif)+relfeature=res101-frcn2018.01 | 84.17 | — | — | |
| PinkEvaluation Setting=Fine-tuning Setting2023.11 | 84 | — | — | |
| AttBalance2025.12 | 83.9 | — | — | |
| SegVG2025.12 | 83.1 | — | — | |
| GroundingGPT (7B)Resolution=336, Visual token count=5762024.03 | 82.47 | — | — | |
| Ferret-7BEvaluation Setting=Fine-tuning Setting2023.11 | 82.45 | — | — | |
| Ferret (7B)Resolution=336, Visual token count=6082024.03 | 82.45 | — | — | |
| Ferret-7BModel Size=7B2024.07 | 82.45 | — | — | |
| MAttNet:subj(+attr)+loc(+dif)+relfeature=vgg162018.01 | 82.3 | — | — | |
| Qwen2-VL-2BType=Generalist2024.09 | 82.3 | — | — | |
| Shikra-13BModel type=Generalist VL SOTAs (w/o finetuning)2023.06 | 81.81 | — | — | |
| Shikra-13BModel type=Generalist Models2023.08 | 81.81 | — | — | |
| Shikra (13B)Model type=Generalist models2023.10 | 81.81 | — | — | |
| Shikra-13BEvaluation Setting=Fine-tuning Setting2023.11 | 81.81 | — | — | |
| Shikra-13BType=Generalist2023.11 | 81.81 | — | — | |
| Shikra (13B)Resolution=224, Visual token count=2562024.03 | 81.81 | — | — | |
| ShikraModel type=Generalist Models, Backbone=Vicuna-13B2023.12 | 81.81 | — | — | |
| Shikra-13BModel Size=13B2024.07 | 81.81 | — | — | |
| Shikra-13BParameters=13B2024.02 | 81.81 | — | — | |
| ExpVG2025.12 | 81.5 | — | — | |
| MDETR2024.07 | 81.41 | — | — | |
| Referring TransformerVisual Features=RN101, Pretrain Images=100k, Multi-task=true2021.06 | 81.16 | — | — | |
| InternVL2-8BType=Generalist2024.09 | 80.7 | — | — |