Knowledge-based Visual Question Answering

Benchmarks

Dataset Name	SOTA Method	Metric
OKVQA	GPT-4o	Accuracy0.885	92	1mo ago
E-VQA Single-Hop	PMSR	Accuracy59.9	52	28d ago
INFOSEEK Unseen Question	DBAgent (SFT: InfoSeek)	Accuracy46.5	42	28d ago
OK-VQA	HinD-CoT-Know	VQA Score68.6	32	2mo ago
OKVQA (val)	SMoLA-PaLI-X_FT (Specialist)	Accuracy66.7	27	3mo ago
KB-VQA Aggregate	InternVL3.5-38B + LoT	Average Score37.5	20	2mo ago
ViQuAE (test)	InternVL3.5-38B + LoT	Overall Accuracy61	20	2mo ago
OVEN (val)	Qwen3-VL-4B	Accuracy (All)25.1	20	2mo ago
INFOSEEK (Unseen Entity)	DBAgent (SFT: InfoSeek)	Accuracy51	19	1mo ago
INFOSEEK	VLM-PRF w/ RL	FR (All)42.5	18	14d ago
E-VQA	LLaVA-1.5	Final Fidelity Rate12.5	18	14d ago
InfoSeek All	DBAgent (SFT: InfoSeek)	Accuracy49.9	16	1mo ago
E-VQA All	DBAgent (SFT: E-VQA)	Accuracy45.8	15	1mo ago
OK-VQA v1.0 (test)	Ours + MCAN	Accuracy61.2	15	3mo ago
Knowledge Benchmarks	JARVIS	Average Score48.2	12	3mo ago
A-OKVQA	EVIAN	Accuracy74.93	9	1mo ago
A-OKVQA (val)	BLIP VQAv2 + SelTDA	Accuracy68.9	8	3mo ago
Infoseek M2KR	PMSR	Accuracy50.5	7	1mo ago
A-OKVQA (test)	BLIP VQAv2	Accuracy59.5	7	3mo ago
OKVQA M2KR	PaLM-E	VQA Score0.661	6	3mo ago
INFOSEEK (Overall)	MMhops-R1	Accuracy33.2	5	3mo ago
A-OKVQA n=1145 (held-out)	Π2	Per-Class Loss0.3805	3	26d ago
E-VQA M2KR	RA-VQAv2 w/ PreFLMR	BEM54.45	3	3mo ago
E-VQA (test)	LamRA	Accuracy56.2	2	3mo ago
Infoseek (test)	RA-VQAv2 w/ PreFLMR	Accuracy30.7	2	3mo ago

Showing 25 of 26 rows