mlsanityData to 7 Oct 2026

GQA: A New Dataset for Real-World Visual Reasoning and Compositional Question Answering

arXiv 1902.09506 ↗First posted 25 Feb 2019cs.CL cs.AI cs.CVcited by 77

Not on the map. Why?

Cited by 34 in Multimodal LLMs / vision-language models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

We introduce GQA, a new dataset for real-world visual reasoning and compositional question answering, seeking to address key shortcomings of previous VQA datasets. We have developed a strong and robust question engine that leverages scene graph structures to create 22M diverse reasoning questions, all come with functional programs that represent their semantics. We use the programs to gain tight control over the answer distribution and present a new tunable smoothing technique to mitigate question biases. Accompanying the dataset is a suite of new metrics that evaluate essential qualities such as consistency, grounding and plausibility. An extensive analysis is performed for baselines as well as state-of-the-art models, providing fine-grained results for different question types and topologies. Whereas a blind LSTM obtains mere 42.1%, and strong VQA models achieve 54.1%, human performance tops at 89.3%, offering ample opportunity for new research to explore. We strongly hope GQA will provide an enabling resource for the next generation of models with enhanced robustness, improved consistency, and deeper semantic understanding for images and language.

Key facts. Cited by 77 papers in the map (data through 7 Oct 2026). First posted on arXiv 25 Feb 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 10; 2022: 17; 2023: 9; 2024: 15; 2025: 15; 2026: 11.

77 papers.

Sort:

By area:

  1. 2021Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Conceptscited by 1,209
  2. 2021InfographicVQAcited by 430
  3. 2023AM-RADIO: Agglomerative Vision Foundation Model -- Reduce All Domains Into Onecited by 165
  4. 2022Vision-Language Pre-training: Basics, Recent Advances, and Future Trendscited by 117
  5. 2021Causal Attention for Vision-Language Taskscited by 111
  6. 2024SAT: Dynamic Spatial Aptitude Training for Multimodal Language Modelscited by 111
  7. 2023Image Captioners Are Scalable Vision Learners Toocited by 89
  8. 2023Getting ViT in Shape: Scaling Laws for Compute-Optimal Model Designcited by 87
  9. 2021MAGMA -- Multimodal Augmentation of Generative Models through Adapter-based Finetuningcited by 85
  10. 2021Multi-Modal Answer Validation for Knowledge-Based VQAcited by 78
  11. 2021Adversarial VQA: A New Benchmark for Evaluating the Robustness of VQA Modelscited by 74
  12. 2022A Survey on Graph Neural Networks and Graph Transformers in Computer Vision: A Task-Oriented Perspectivecited by 56
38 more
  1. 2024From Multimodal LLMs to Generalist Embodied Agents: Methods and Lessonscited by 42
  2. 2024EXAMS-V: A Multi-Discipline Multilingual Multimodal Exam Benchmark for Evaluating Vision Language Modelscited by 41
  3. 2022Visual Commonsense in Pretrained Unimodal and Multimodal Modelscited by 39
  4. 2024LocCa: Visual Pretraining with Location-aware Captionerscited by 35
  5. 2023DesCo: Learning Object Recognition with Rich Language Descriptionscited by 25
  6. 2022There is a Time and Place for Reasoning Beyond the Imagecited by 22
  7. 2021ICDAR 2021 Competition on Document VisualQuestion Answeringcited by 19
  8. 2024VALOR-EVAL: Holistic Coverage and Faithfulness Evaluation of Large Vision-Language Modelscited by 16
  9. 2021A First Look: Towards Explainable TextVQA Models via Visual and Textual Explanationscited by 14
  10. 2022Multi-modal Misinformation Detection: Approaches, Challenges and Opportunitiescited by 14
  11. 2025VisiPruner: Decoding Discontinuous Cross-Modal Dynamics for Efficient Multimodal LLMscited by 14
  12. 2022FindIt: Generalized Localization with Natural Language Queriescited by 13
  13. 2022CRIPP-VQA: Counterfactual Reasoning about Implicit Physical Properties via Video Question Answeringcited by 13
  14. 2022Answer-Me: Multi-Task Open-Vocabulary Visual Question Answeringcited by 12
  15. 2024UNIT: Unifying Image and Text Recognition in One Vision Encodercited by 12
  16. 2024Typhoon 2: A Family of Open Text and Multimodal Thai Large Language Modelscited by 12
  17. 2022Pre-training image-language transformers for open-vocabulary taskscited by 11
  18. 2025Robust Adaptation of Large Multimodal Models for Retrieval Augmented Hateful Meme Detectioncited by 11
  19. 2025SciVideoBench: Benchmarking Scientific Video Reasoning in Large Multimodal Modelscited by 11
  20. 2022Consistency-preserving Visual Question Answering in Medical Imagingcited by 10
  21. 2023Effective End-to-End Vision Language Pretraining with Semantic Visual Losscited by 10
  22. 2023A Study of Autoregressive Decoders for Multi-Tasking in Computer Visioncited by 10
  23. 2025From Illusion to Intention: Visual Rationale Learning for Vision-Language Reasoningcited by 9
  24. 2022Dynamic Key-value Memory Enhanced Multi-step Graph Reasoning for Knowledge-based Visual Question Answeringcited by 7
  25. 2022Entity-Focused Dense Passage Retrieval for Outside-Knowledge Visual Question Answeringcited by 7
  26. 2022ERNIE-UniX2: A Unified Cross-lingual Cross-modal Framework for Understanding and Generationcited by 7
  27. 2024FlexCap: Describe Anything in Images in Controllable Detailcited by 7
  28. 2024Relational Programming with Foundation Modelscited by 7
  29. 2025Conformal Prediction and MLLM aided Uncertainty Quantification in Scene Graph Generationcited by 7
  30. 2022Self-Training Vision Language BERTs with a Unified Conditional Modelcited by 6
  31. 2021Topos and Stacks of Deep Neural Networkscited by 5
  32. 2023Localized Questions in Medical Visual Question Answeringcited by 5
  33. 2025SemVink: Advancing VLMs' Semantic Understanding of Optical Illusions via Visual Global Thinkingcited by 5
  34. 2026From Seeing to Thinking: Decoupling Perception and Reasoning Improves Post-Training of Vision-Language Modelscited by 5
  35. 2024RTGen: Generating Region-Text Pairs for Open-Vocabulary Object Detectioncited by 4
  36. 2025TinyAlign: Boosting Lightweight Vision-Language Models by Mitigating Modal Alignment Bottleneckscited by 4
  37. 2025Caption This, Reason That: VLMs Caught in the Middlecited by 4
  38. 2025Enhancing Vision-Language Model Training with Reinforcement Learning in Synthetic Worlds for Real-World Successcited by 4

50 most cited of 77 All 77 (JSON)