HorusEye：语言作为动态注意力用于应急视觉分析
HorusEye: Language as Dynamic Attention for Emergency Visual Analysis
AI总结：
提出HorusEye框架，通过语言反馈动态引导视觉分析，在应急场景下评估多种VLM，发现语言反馈效果依赖模型，并揭示热成像中的裁剪悖论。
摘要：
我们介绍了HorusEye，即语言作为动态注意力用于应急视觉分析。我们的研究分为五个阶段。第一阶段是构建RefCOCO-Degraded基准数据集，包含15,244张图像（3,811张基础图像×4种条件：清晰、雾、烟和热成像），具有系统性的视觉退化。通过四个研究问题，我们评估了多种VLM（Gemini、Qwen2-VL、BLIP-2、LLaVA、Kosmos-2）在视觉定位（第二阶段）、语言反馈恢复（第三阶段）、健康VQA任务（第四阶段）以及幻觉分析（最终阶段）上的表现。我们的关键发现是语言反馈的有效性依赖于模型：Gemini通过迭代语言反馈在热成像条件下提升了47.3%，而Qwen2-VL在相同协议下性能下降了5.1%。我们还发现了“热成像悖论”，即提升RGB性能的裁剪策略在热成像中灾难性地失败。此外，BLIP-2在退化条件下独特地产生更多幻觉，使其不适合应急部署。
英文摘要：
We introduce HorusEye, Language as Dynamic Attention for Emergency Visual Analysis. Our investigation followed five stages. The first one is benchmarking RefCOCO-Degraded, a dataset of 15,244 images (3,811 base images x 4 conditions: Clean, Fog, Smoke and Thermal) with systematic visual degradation. Through four research questions, we evaluate multiple VLMs (Gemini, Qwen2-VL, BLIP-2, LLaVA, Kosmos-2) across visual grounding the second stage, language feedback recovery the third one, health VQA tasks the fourth, and hallucination analysis the final stage. Our key finding is that language feedback effectiveness is model-dependent: Gemini achieves +47.3% improvement in thermal conditions through iterative language feedback, while Qwen2-VL shows -5.1% degradation under the same protocol. We also identify the 'Thermal Paradox' where cropping strategies that improve RGB performance catastrophically fail in thermal imagery. Furthermore, BLIP-2 uniquely hallucinates more under degradation, making it unsuitable for emergency deployment
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。
