研究人员推出DetAS，一个将目标检测视为动态决策过程的agentic框架。该框架利用多模态大语言模型（MLLM）通过从恢复模块和专用检测器的工具箱中进行选择，自适应地组合检测工作流。扩展的DetAS-X版本通过累积带注释数据的经验来进一步优化决策质量，使其能够在推理过程中逐步调整其策略。实验表明，DetAS-X的性能显著优于现有的基于MLLM的检测器，在具有挑战性的基准测试中取得了显著的F1分数提升。 AI
影响 引入了一种新颖的目标检测agentic方法，有可能在复杂、动态的环境中提高性能。
排序理由 该集群包含一篇详细介绍目标检测新框架的研究论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.31174v1 Announce Type: cross Abstract: Object detection in real-world scenarios remains challenging due to diverse image degradations and heterogeneous object distributions, which significantly hinder the generalization of existing detectors. Conventional approaches, i…
Object detection in real-world scenarios remains challenging due to diverse image degradations and heterogeneous object distributions, which significantly hinder the generalization of existing detectors. Conventional approaches, including scene-specific representation learning an…