眼见为实：通过反向视觉定位实现多模态大语言模型的富上下文幻觉检测
Seeing is Believing: Rich-Context Hallucination Detection for MLLMs via Backward Visual Grounding
AI总结：
针对多模态大语言模型（MLLMs）的幻觉问题，提出无参考检测框架VBackChecker，通过像素级定位大语言模型验证响应与视觉输入的一致性，设计R-Instruct数据生成 pipeline 并构建R²-HalBench基准，在相关任务上实现最先进性能。
摘要：
多模态大语言模型（MLLMs）已解锁强大的跨模态能力，但仍受幻觉问题严重困扰。因此，准确检测MLLMs的幻觉对确保其在实际应用中的可靠性至关重要。为此，遵循“眼见为实”原则，我们引入VBackChecker——一个新颖的无参考幻觉检测框架，通过利用具备推理和指代分割能力的像素级定位大语言模型（Grounding LLM），验证MLLM生成的响应与视觉输入的一致性。该无参考框架不仅有效处理富上下文场景，还提供可解释性。为推进这一工作，我们相应设计了创新的指令微调数据生成 pipeline（R-Instruct），其特点是包含富上下文描述、定位掩码和难负样本。我们进一步构建了R²-HalBench——一个新的MLLM幻觉基准，与现有基准不同，它涵盖来自18个MLLM的真实世界富上下文描述，包含多样的对象、属性和关系级细节。VBackChecker优于此前的复杂框架，在R²-HalBench上实现了最先进性能，甚至可与GPT-4o的幻觉检测能力相媲美；它在像素级定位任务中也超过现有方法，实现了10%以上的提升。所有代码、数据和模型可在https://github.com/PinxueGuo/VBackChecker获取。
英文摘要：
Multimodal Large Language Models (MLLMs) have unlocked powerful cross-modal capabilities, but still significantly suffer from hallucinations. As such, accurate detection of hallucinations in MLLMs is imperative for ensuring their reliability in practical applications. To this end, guided by the principle of "Seeing is Believing", we introduce VBackChecker, a novel reference-free hallucination detection framework that verifies the consistency of MLLMgenerated responses with visual inputs, by leveraging a pixellevel Grounding LLM equipped with reasoning and referring segmentation capabilities. This reference-free framework not only effectively handles rich-context scenarios, but also offers interpretability. To facilitate this, an innovative pipeline is accordingly designed for generating instruction-tuning data (R-Instruct), featuring rich-context descriptions, grounding masks, and hard negative samples. We further establish R^2 -HalBench, a new hallucination benchmark for MLLMs, which, unlike previous benchmarks, encompasses real-world, rich-context descriptions from 18 MLLMs with high-quality annotations, spanning diverse object-, attribute, and relationship-level details. VBackChecker outperforms prior complex frameworks and achieves state-of-the-art performance on R^2 -HalBench, even rivaling GPT-4o's capabilities in hallucination detection. It also surpasses prior methods in the pixel-level grounding task, achieving over a 10% improvement. All codes, data, and models are available at https://github.com/PinxueGuo/VBackChecker.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
