Papers
Topics
Authors
Recent
Search
2000 character limit reached

Don't Miss the Forest for the Trees: Attentional Vision Calibration for Large Vision Language Models

Published 28 May 2024 in cs.CV and cs.AI | (2405.17820v2)

Abstract: Large Vision LLMs (LVLMs) demonstrate strong capabilities in visual understanding and description, yet often suffer from hallucinations, attributing incorrect or misleading features to images. We observe that LVLMs disproportionately focus on a small subset of image tokens--termed blind tokens--which are typically irrelevant to the query (e.g., background or non-object regions). We hypothesize that such attention misalignment plays a key role in generating hallucinated responses. To mitigate this issue, we propose Attentional Vision Calibration (AvisC), a test-time approach that dynamically recalibrates the influence of blind tokens without modifying the underlying attention mechanism. AvisC first identifies blind tokens by analyzing layer-wise attention distributions over image tokens, then employs a contrastive decoding strategy to balance the influence of original and blind-token-biased logits. Experiments on standard benchmarks, including POPE, MME, and AMBER, demonstrate that AvisC effectively reduces hallucinations in LVLMs.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (50)
  1. Qwen-vl: A versatile vision-language model for understanding, localization, text reading, and beyond. 2023.
  2. Mocha: Multi-objective reinforcement mitigating caption hallucinations. arXiv preprint arXiv:2312.03631, 2023.
  3. End-to-end object detection with transformers. In European conference on computer vision, pages 213–229. Springer, 2020.
  4. Shikra: Unleashing multimodal llm’s referential dialogue magic. arXiv preprint arXiv:2306.15195, 2023.
  5. Halc: Object hallucination reduction via adaptive focal-contrast decoding. arXiv preprint arXiv:2403.00425, 2024.
  6. Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks. arXiv preprint arXiv:2312.14238, 2023.
  7. Mitigating hallucination in visual language models with visual supervision. arXiv preprint arXiv:2311.16479, 2023.
  8. Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality. See https://vicuna. lmsys. org (accessed 14 April 2023), 2(3):6, 2023.
  9. Instructblip: Towards general-purpose vision-language models with instruction tuning. Advances in Neural Information Processing Systems, 36, 2024.
  10. Seeing is believing: Mitigating hallucination in large vision-language models via clip-guided decoding. arXiv preprint arXiv:2402.15300, 2024.
  11. Multi-modal hallucination control by visual information grounding. arXiv preprint arXiv:2403.14003, 2024.
  12. Mme: A comprehensive evaluation benchmark for multimodal large language models. arXiv preprint arXiv:2306.13394, 2024.
  13. Detecting and preventing hallucinations in large vision language models. arXiv preprint arXiv:2308.06394, 2023.
  14. Detecting and preventing hallucinations in large vision language models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 18135–18143, 2024.
  15. Incorporating visual experts to resolve the information loss in multimodal large language models. arXiv preprint arXiv:2401.03105, 2024.
  16. Opera: Alleviating hallucination in multi-modal large language models via over-trust penalty and retrospection-allocation. arXiv preprint arXiv:2311.17911, 2023.
  17. Gqa: A new dataset for real-world visual reasoning and compositional question answering. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6700–6709, 2019.
  18. Vcoder: Versatile vision encoders for multimodal large language models. arXiv preprint arXiv:2312.14233, 2023.
  19. Hallucination augmented contrastive learning for multimodal large language model. arXiv preprint arXiv:2312.06968, 2023.
  20. Volcano: mitigating multimodal hallucination through self-feedback guided revision. arXiv preprint arXiv:2311.07362, 2023.
  21. Mitigating object hallucinations in large vision-language models through visual contrastive decoding. arXiv preprint arXiv:2311.16922, 2023.
  22. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International conference on machine learning, pages 19730–19742. PMLR, 2023.
  23. Visual evidence prompting mitigates hallucinations in multimodal large language models. 2023.
  24. Evaluating object hallucination in large vision-language models. arXiv preprint arXiv:2305.10355, 2023.
  25. Microsoft coco: Common objects in context. In Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13, pages 740–755. Springer, 2014.
  26. Mitigating hallucination in large multi-modal models via robust instruction tuning. In The Twelfth International Conference on Learning Representations, 2023.
  27. Improved baselines with visual instruction tuning. arXiv preprint arXiv:2310.03744, 2023.
  28. Visual instruction tuning. Advances in neural information processing systems, 36, 2023.
  29. Mmbench: Is your multi-modal model an all-around player? arXiv preprint arXiv:2307.06281, 2023.
  30. Evaluation and enhancement of semantic grounding in large vision-language models. In AAAI-ReLM Workshop, 2024.
  31. Learning transferable visual models from natural language supervision. In International conference on machine learning, pages 8748–8763. PMLR, 2021.
  32. Object hallucination in image captioning. arXiv preprint arXiv:1809.02156, 2018.
  33. A-okvqa: A benchmark for visual question answering using world knowledge. In European Conference on Computer Vision, pages 146–162. Springer, 2022.
  34. Aligning large multimodal models with factually augmented rlhf. arXiv preprint arXiv:2309.14525, 2023.
  35. Eyes wide shut? exploring the visual shortcomings of multimodal llms. arXiv preprint arXiv:2401.06209, 2024.
  36. Contrastive region guidance: Improving grounding in vision-language models without training. arXiv preprint arXiv:2403.02325, 2024.
  37. Vigc: Visual instruction generation and correction. arXiv preprint arXiv:2308.12714, 2023.
  38. Amber: An llm-free multi-dimensional benchmark for mllms hallucination evaluation. arXiv preprint arXiv:2311.07397, 2023.
  39. Mitigating hallucinations in large vision-language models with instruction contrastive decoding. arXiv preprint arXiv:2403.18715, 2024.
  40. Logical closed loop: Uncovering object hallucinations in large vision-language models. arXiv preprint arXiv:2402.11622, 2024.
  41. Pensieve: Retrospect-then-compare mitigates visual hallucination. arXiv preprint arXiv:2403.14401, 2024.
  42. Hallucidoctor: Mitigating hallucinatory toxicity in visual instruction data. arXiv preprint arXiv:2311.13614, 2023.
  43. Rlhf-v: Towards trustworthy mllms via behavior alignment from fine-grained correctional human feedback. arXiv preprint arXiv:2312.00849, 2023.
  44. Less is more: Mitigating multimodal hallucination from an eos decision perspective. arXiv preprint arXiv:2402.14545, 2024.
  45. Halle-switch: Controlling object hallucination in large vision language models. arXiv e-prints, pages arXiv–2310, 2023.
  46. Debiasing large visual language models. arXiv preprint arXiv:2403.05262, 2024.
  47. Mitigating object hallucination in large vision-language models via classifier-free guidance. arXiv preprint arXiv:2402.08680, 2024.
  48. Beyond hallucinations: Enhancing lvlms through hallucination-aware direct preference optimization. arXiv preprint arXiv:2311.16839, 2023.
  49. Analyzing and mitigating object hallucination in large vision-language models. arXiv preprint arXiv:2310.00754, 2023.
  50. Minigpt-4: Enhancing vision-language understanding with advanced large language models. arXiv preprint arXiv:2304.10592, 2023.
Citations (3)
View on Semantic Scholar

Summary

No one has generated a summary of this paper yet.

Sign Up to Summarize

Paper to Video (Beta)

No one has generated a video about this paper yet.

Sign Up to Generate All Videos Subscribe on YouTube

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Sign Up to Generate

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Generate Now

Continue Learning

We haven't generated follow-up questions for this paper yet.

Generate Now

Collections

Sign up for free to add this paper to one or more collections.

Sign Up

Tweets

Sign up for free to view the 1 tweet with 0 likes about this paper.

Sign Up for Free