List your company
ResearchFormerly WizWand

External Knowledge-dependent Image Question Answering on OK-VQA

91.9AccuracyStaR-KVQA_Gemma
11.851232.633153.41574.1969Mar 14, 2023Aug 29, 2023Feb 13, 2024Jul 31, 2024Jan 15, 2025Jul 2, 2025Dec 18, 2025
Updated 7mo ago

Evaluation results

MethodLinks
StaR-KVQA_GemmaModel Inputs=Question + Image, External Knowledge=Fine-tuned Gemma-3-12B2025.1091.9
StaR-KVQA_QwenModel Inputs=Question + Image, External Knowledge=Fine-tuned Qwen2.5-VL-7B2025.1091.51
StaR-KVQA_LlamaModel Inputs=Question + Image, External Knowledge=Fine-tuned Llama-3.2-11B-Vision2025.1090.01
SDFTModel Inputs=Question + Image, External Knowledge=Fine-tuned Qwen2.5-VL-7B2025.1082.56
Qwen2.5-VL-72BModel Inputs=Question + Image, External Knowledge=Qwen2.5-VL-72B2025.1080.75
Gemini 2.5 ProModel Inputs=Question + Image, External Knowledge=Gemini 2.5 Pro2025.1080.53
Gemini 2.5 FlashModel Inputs=Question + Image, External Knowledge=Gemini 2.5 Flash2025.1079.97
CoT + SFTModel Inputs=Question + Image, External Knowledge=Fine-tuned Qwen2.5-VL-7B2025.1079.58
Gemma-3-27BModel Inputs=Question + Image, External Knowledge=Gemma-3-27B2025.1079.34
M2-ReasoningModel Inputs=Question + Image, External Knowledge=M2-Reasoning-7B2025.1078.63
GPT-4oModel Inputs=Question + Image, External Knowledge=GPT-4o2025.1077.86
CoTModel Inputs=Question + Image, External Knowledge=Qwen2.5-VL-7B2025.1076.88
LLaVA-CoTModel Inputs=Question + Image, External Knowledge=Fine-tuned Llama-3.2-11B-Vision2025.1076.57
SFTModel Inputs=Question + Image, External Knowledge=Fine-tuned Qwen2.5-VL-7B2025.1076.36
Qwen2.5-VL-7BModel Inputs=Question + Image, External Knowledge=Qwen2.5-VL-7B2025.1075.74
Gemma-3-12BModel Inputs=Question + Image, External Knowledge=Gemma-3-12B2025.1071.4
PaliGemma2-28BResolution=448x448, Fine-tuning=per-task2025.1270.6
PaliGemma2-3B + AuditDMResolution=448x448, Fine-tuning=per-task2025.1269.2
PaliGemma2-10BResolution=448x448, Fine-tuning=per-task2025.1268.6
Llama-3.2-11B-VisionModel Inputs=Question + Image, External Knowledge=Llama-3.2-11B-Vision2025.1067.84
InternVL3-78BModel Inputs=Question + Image, External Knowledge=InternVL3-78B2025.1067.61
PaliGemma2-3BResolution=448x448, Fine-tuning=per-task2025.1264.1
PromptCapEvaluation Protocol=Supervised2023.0358.8
REVIVEEvaluation Protocol=Supervised2023.0358
MAILModel Inputs=Question + Image, External Knowledge=Frozen MiniGPT-4 (7B) + ConceptNet2025.1056.69
RA-VQAEvaluation Protocol=Supervised2023.0354.5
KAT (Ensemble)Model Inputs=Question + Caption + Object Tags, External Knowledge=Frozen GPT-3 (175B) + Wikidata2025.1054.41
KATEvaluation Protocol=Supervised2023.0354.4
REVIVEModel Inputs=Question + Caption + Region Tags, External Knowledge=Frozen GPT-3 (175B) + Wikidata2025.1053.83
KAT (Single)Model Inputs=Question + Caption + Object Tags, External Knowledge=Frozen GPT-3 (175B) + Wikidata2025.1053.09
ViperGPTEvaluation Protocol=Zero-shot2023.0351.9
FlamingoEvaluation Protocol=Zero-shot2023.0350.6
TRIGEvaluation Protocol=Supervised2023.0350.5
Pica-FullModel Inputs=Question + Caption + Object Tags, External Knowledge=Frozen GPT-3 (175B)2025.1048
BLIP-2Evaluation Protocol=Zero-shot2023.0345.9
MCANModel Inputs=Question + Image, External Knowledge=-2025.1044.65
PICaEvaluation Protocol=Zero-shot2023.0343.3
PICA-BaseModel Inputs=Question + Caption + Object Tags, External Knowledge=Frozen GPT-3 (175B)2025.1043.3
VLC-BERTModel Inputs=Question + Image, External Knowledge=COMET + ConceptNet2025.1043.14
MAVExModel Inputs=Question + Image, External Knowledge=Wikipedia + ConceptNet + Google Images2025.1041.37
KrispModel Inputs=Question + Image, External Knowledge=Wikipedia + ConceptNet2025.1038.9
HCNMNModel Inputs=Question + Image, External Knowledge=WordNet2025.1036.74
PNP-VQAEvaluation Protocol=Zero-shot2023.0335.9
ConceptBERTModel Inputs=Question + Image, External Knowledge=ConceptNet2025.1033.66
MUTAN +ANModel Inputs=Question + Image, External Knowledge=Wikipedia2025.1027.84
MUTANModel Inputs=Question + Image, External Knowledge=-2025.1026.41
BAN +ANModel Inputs=Question + Image, External Knowledge=Wikipedia2025.1025.61
BANModel Inputs=Question + Image, External Knowledge=-2025.1025.17
Q OnlyModel Inputs=Question + Image, External Knowledge=-2025.1014.93