Share your thoughts, 1 month free Claude Pro on usSee more

Visual Grounding on RefCOCOg (test)

93.3Accuracy

Vanilla

Updated 5d ago

Evaluation Results

Method
Vanilla 2026.02	93.3	-	-	-	-
Youtu-VL 2026.01	92.9	-	-	-	-
IVC-Prune 2026.02	92.4	-	-	-	-
InternVL2.5 2025.02	92.2	-	-	-	-
Florence-2 2026.01	91.7	-	-	-	-
PDrop 2026.02	91.5	-	-	-	-
InternVL-3-78B 2026.01	91.5	-	-	-	-
W-FastV 2026.02	91.3	-	-	-	-
Qwen3-VL-8B-Thinking + EGM 2026.01	91.2	-	-	-	-
Qwen3-VL-4B-Thinking + EGM 2026.01	90.8	-	-	-	-
Qwen3-VL-235B-A22B-Instruct 2026.01	90.8	-	-	-	-
CogVLM-Grounding-17B 2024.02	90.79	-	-	-	-
InternVL-3-8B + EGM 2026.01	90.7	-	-	-	-
FastV 2026.02	90.6	-	-	-	-
Qwen3-VL-235B-A22B-Thinking 2026.01	90.5	-	-	-	-
Qwen2.5-VL 2025.02	90.3	-	-	-	-
Original Grounding CogVLM 2024.06	90.2	-	-	0	-
InternVL-3-8B 2026.01	90	-	-	-	-
InfMLLM-7B 2023.11	89.78	-	-	-	-
UNINEXT 2024.03	89.73	-	-	-	-
InternVL-3.5 2026.01	89.3	-	-	-	-
Qwen3-VL-2B-Thinking + EGM 2026.01	89.3	-	-	-	-
UFO 2026.01	89.2	-	-	-	-
Non-merging CogVLM 2024.06	88.9	-	-	0	-
Qwen3-VL-8B-Instruct 2026.01	88.7	-	-	-	-
InternVL-3-2B + EGM 2026.01	88.7	-	-	-	-
Qwen3-VL-8B-Thinking 2026.01	88.6	-	-	-	-
Qwen3-VL-4B-Instruct 2026.01	87.8	-	-	-	-
HPE-CogVLM 2024.06	87.8	-	-	0	-
Qwen3-VL 2026.01	87.7	-	-	-	-
Qwen3-VL-4B-Thinking 2026.01	87.7	-	-	-	-
Vanilla 2026.02	87.6	-	-	-	-
InternVL-3-1B + EGM 2026.01	87.5	-	-	-	-
Vanilla 2026.02	87.24	-	-	-	-
Qwen2.5-VL 2025.02	87.2	-	-	-	-
Griffon 2026.01	87.2	-	-	-	-
InternVL-3-2B 2026.01	87.2	-	-	-	-
G-DINO-L 2024.03	87.02	-	-	-	-
Grounding DINO 2025.02	87	-	-	-	-
Grounding DINO 2026.01	87	-	-	-	-
Vanilla 2026.02	86.9	-	-	-	-
IVC-Prune 2026.02	86.9	-	-	-	-
Nuwa 2026.02	86.87	-	-	-	-
IVC-Prune 2026.02	86.5	-	-	-	-
TA merging CogVLM 2024.06	86.1	-	-	0	-
Qwen2.5-VL 2025.02	85.7	-	-	-	-
Qwen3-VL-2B-Instruct 2026.01	85.7	-	-	-	-
Nuwa 2026.02	85.49	-	-	-	-
QWen-VL-7B 2023.11	85.48	-	-	-	-
Qwen-VL-7B 2024.02	85.48	-	-	-	-
STORM 2026.04	85.1	-	-	-	-
Ferret 2026.04	84.8	-	-	-	-
Ferret-7B 2024.02	84.76	-	-	-	-
Individual VQA 2025.05	84.72	-	-	-	-
M-GPT2 2026.04	84.7	-	-	-	-
OptMerge 2025.05	84.68	-	-	-	-
MiniGPT-v2-7B 2023.11	84.66	-	-	-	-
Individual OCR 2025.05	84.62	-	-	-	-
Individual Geometry 2025.05	84.56	-	-	-	-
Qwen3-VL-2B-Thinking 2026.01	84.5	-	-	-	-
Lumen-7B 2024.03	84.44	-	-	-	-
VistaLLM-7B 2024.02	84.4	-	-	-	-
Individual Chart 2025.05	84.33	-	-	-	-
VisionLLM v2 2026.01	84.1	-	-	-	-
Qwen2.5-VL-32B-Instruct 2025.05	84.06	-	-	-	-
Individual Grounding 2025.05	84.04	-	-	-	-
VScan 2026.02	83.6	-	-	-	-
MDETR 2026.01	83.3	-	-	-	-
Shik-13B 2026.04	83.2	-	-	-	-
Shikra-13B 2023.11	83.16	-	-	-	-
Shikra 2026.02	83.1	-	-	-	-
InternVL-3-1B 2026.01	82.6	-	-	-	-
CoLLaVO-7B 2024.02	82.33	-	-	-	-
Shik-7B 2026.04	82.2	-	-	-	-
Shikra-7B 2023.11	82.19	-	-	-	-
Shikra-7B 2024.03	82.19	-	-	-	-
Shikra-7B 2024.02	82.19	-	-	-	-
G-GPT 2026.04	82	-	-	-	-
Fully-FT 2023.08	80.95	-	-	-	-
Fully-FT 2026.04	80.95	-	-	-	-
Partially⇑ 2023.08	80.08	-	-	-	-
W-FastV 2026.02	79.2	-	-	-	-
Ours♠ 2026.04	78.13	-	-	-	-
FastV 2026.02	77.9	-	-	-	-
Ours♡ 2026.04	77.73	-	-	-	-
PDrop 2026.02	77.7	-	-	-	-
UniPT 2023.08	77.33	-	-	-	-
UniPT 2026.04	77.33	-	-	-	-
SHERL 2026.04	77.33	-	-	-	-
Gemini 1.5 Pro 2025.02	76.2	-	-	-	-
MMnasNet 2020.04	75.7	-	-	-	-
FastV 2026.02	75.3	-	-	-	-
GLaMM 2026.01	74.9	-	-	-	-
W-FastV 2026.02	74.4	-	-	-	-
Nuwa 2026.02	73.96	-	-	-	-
LST 2023.08	73.67	-	-	-	-
LST 2026.04	73.67	-	-	-	-
MMnasNet 2020.04	72.7	-	-	-	-
PFOS 2026.02	67.6	-	-	-	-
OFA-L 2023.11	67.58	-	-	-	-

Showing 100 of 176 rows