Visual Question Answering (VQA)

Visual Question Answering (VQA) is a task in computer vision that involves answering questions about an image. The goal of VQA is to teach machines to understand the content of an image and answer questions about it in natural language.

Image Source: visualqa.org

Papers

Recently Added Most Hyped Most Active Needs Verification Most Verified

Showing 1–10 of 2167 papers

Title	Date	Tasks	Status	Hype
Qwen2.5-VL Technical Report	Feb 19, 2025	document understanding	CodeCode Available	11
Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution	Sep 18, 2024	Natural Language Visual Grounding	CodeCode Available	11
SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning	Aug 10, 2024	HallucinationOptical Character Recognition	CodeCode Available	11
Infinity-MM: Scaling Multimodal Performance with Large-Scale and High-Quality Instruction Data	Oct 24, 2024	Image GenerationQuestion Generation	CodeCode Available	7
mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models	Aug 9, 2024	Language ModelingLanguage Modelling	CodeCode Available	7
MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models	Apr 20, 2023	Image DescriptionLanguage Modelling	CodeCode Available	7
Improved Baselines with Visual Instruction Tuning	Oct 5, 2023	Factual Inconsistency Detection in Chart CaptioningImage Classification	CodeCode Available	6
GPT-4 Technical Report	Mar 15, 2023	answerability predictionArithmetic Reasoning	CodeCode Available	6
VisionLLM v2: An End-to-End Generalist Multimodal Large Language Model for Hundreds of Vision-Language Tasks	Jun 12, 2024	Image GenerationLanguage Modeling	CodeCode Available	5
VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs	Jun 11, 2024	Multiple-choiceQuestion Answering	CodeCode Available	5

Show:10 25 50

← PrevPage 1 of 217Next →

All datasets GQA Test2019 VQA v2 test-dev VQA v2 test-std OK-VQA MSVD-QA DocVQA test MSRVTT-QA InfographicVQA GQA test-dev VizWiz 2020 VQA A-OKVQA CLEVR

Benchmark Results

#	Model	Metric	Claimed	Verified	Status
1	human	Accuracy	89.3	—	Unverified
2	DREAM+Unicoder-VL (MSRA)	Accuracy	76.04	—	Unverified
3	TRRNet (Ensemble)	Accuracy	74.03	—	Unverified
4	MIL-nbgao	Accuracy	73.81	—	Unverified
5	Kakao Brain	Accuracy	73.33	—	Unverified
6	Coarse-to-Fine Reasoning, Single Model	Accuracy	72.14	—	Unverified
7	270	Accuracy	70.23	—	Unverified
8	NSM ensemble (updated)	Accuracy	67.55	—	Unverified
9	VinVL-DPT	Accuracy	64.92	—	Unverified
10	VinVL+L	Accuracy	64.85	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	PaLI	Accuracy	84.3	—	Unverified
2	BEiT-3	Accuracy	84.19	—	Unverified
3	VLMo	Accuracy	82.78	—	Unverified
4	ONE-PEACE	Accuracy	82.6	—	Unverified
5	mPLUG (Huge)	Accuracy	82.43	—	Unverified
6	CuMo-7B	Accuracy	82.2	—	Unverified
7	X2-VLM (large)	Accuracy	81.9	—	Unverified
8	MMU	Accuracy	81.26	—	Unverified
9	InternVL-C	Accuracy	81.2	—	Unverified
10	Lyrics	Accuracy	81.2	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	BEiT-3	overall	84.03	—	Unverified
2	mPLUG-Huge	overall	83.62	—	Unverified
3	ONE-PEACE	overall	82.52	—	Unverified
4	X2-VLM (large)	overall	81.8	—	Unverified
5	VLMo	overall	81.3	—	Unverified
6	SimVLM	overall	80.34	—	Unverified
7	X2-VLM (base)	overall	80.2	—	Unverified
8	VAST	overall	80.19	—	Unverified
9	VALOR	overall	78.62	—	Unverified
10	Prompt Tuning	overall	78.53	—	Unverified