提出DASH框架，自动发现视觉语言模型在真实图像中的系统性幻觉。
DASH: Detection and Assessment of Systematic Hallucinations of VLMs
- 基于自然图像流形优化生成误导性图像，检测系统性幻觉
- 在380类物体上发现超19,000个幻觉集群，共95万张图像
- 可迁移至其他模型，微调后显著降低幻觉率
视觉语言模型（VLMs）容易产生物体幻觉，即错误地声称图像中存在某些物体。现有基准依赖小规模标注数据集评估幻觉，难以覆盖开放世界场景下的系统性错误。本文提出DASH（Detection and Assessment of Systematic Hallucinations），一个自动化的大型管道，用于在真实图像的开放世界设置下识别VLM的系统性幻觉。核心组件DASH-OPT通过在“自然图像流形”上优化，生成能误导VLM的图像。DASH输出为包含真实且语义相似图像的聚类，其中VLM会幻觉出特定物体。我们在PaliGemma和两个LLaVA-NeXT模型上对380个物体类别进行测试，共发现超过19,000个聚类，包含95万张图像。我们研究了幻觉在其他VLM间的迁移性，并表明用DASH生成的特定模型图像对PaliGemma进行微调，可有效缓解物体幻觉。代码与数据已公开于https://YanNeu.github.io/DASH。
原文摘要 · Abstract (English)
Vision-language models (VLMs) are prone to object hallucinations, where they erroneously indicate the presenceof certain objects in an image. Existing benchmarks quantify hallucinations using relatively small, labeled datasets. However, this approach is i) insufficient to assess hallucinations that arise in open-world settings, where VLMs are widely used, and ii) inadequate for detecting systematic errors in VLMs. We propose DASH (Detection and Assessment of Systematic Hallucinations), an automatic, large-scale pipeline designed to identify systematic hallucinations of VLMs on real-world images in an open-world setting. A key component is DASH-OPT for image-based retrieval, where we optimize over the ''natural image manifold'' to generate images that mislead the VLM. The output of DASH consists of clusters of real and semantically similar images for which the VLM hallucinates an object. We apply DASH to PaliGemma and two LLaVA-NeXT models across 380 object classes and, in total, find more than 19k clusters with 950k images. We study the transfer of the identified systematic hallucinations to other VLMs and show that fine-tuning PaliGemma with the model-specific images obtained with DASH mitigates object hallucinations. Code and data are available at https://YanNeu.github.io/DASH.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。