边建图边思考：用于增量式3D场景图的异步视觉-语言智能体
Think While You Map: Asynchronous Vision-Language Agents for Incremental 3D Scene Graphs
AI总结：
提出异步架构，轻量在线建图与重型语义精化并发运行，通过语义闭环、属性标注和空间关系推导，实现探索时可查询且语义渐增的3D场景图，在多个基准上超越现有方法。
摘要：
开放词汇的3D场景图方法通常分两阶段运行：先重建，然后用视觉-语言模型丰富，导致探索期间图不可查询。我们认为这种顺序耦合是不必要的，并提出一种异步架构，其中轻量在线建图与重型语义精化并发运行。基于概率体素的骨干网络增量式维护稳定的对象身份，而后台VLM智能体逐步丰富图。该框架通过语义闭环解决重复对象轨迹，附加细粒度视觉属性，并推导对象间的空间关系。多目标帧调度器通过选择少量信息丰富的帧共同覆盖多个目标来分摊VLM成本。生成的场景图在探索期间可查询，并随时间增长语义丰富度。我们的方法在语义分割（ScanNet, Replica）上匹配或超越现有开放词汇3D场景图方法，并在三个视觉定位基准（Sr3D+, Nr3D, ScanRefer）上以A@0.25指标超过先前最先进方法15.3至18.8。项目页面：此 https URL
英文摘要：
Open-vocabulary 3D scene graph methods typically operate in two stages: first reconstruct, then enrich with vision-language models, leaving the graph unqueryable during exploration. We argue that this sequential coupling is unnecessary and propose an asynchronous architecture in which lightweight online mapping runs concurrently with heavyweight semantic refinement. A probabilistic voxel-based backbone maintains stable object identities incrementally, while background VLM agents progressively enrich the graph. This framework resolves duplicate object tracks through semantic loop closure, attaches fine-grained visual attributes and derives spatial relations between objects. A multi-target frame scheduler amortizes VLM cost by selecting a small set of informative frames that jointly cover multiple targets. The resulting scene graph is queryable during exploration and grows in semantic richness over time. Our method matches or outperforms existing open-vocabulary 3D scene graph methods on semantic segmentation (ScanNet, Replica) and surpasses the prior state-of-the-art across three visual grounding benchmarks (Sr3D+, Nr3D, ScanRefer) by 15.3 to 18.8 A@0.25. Project page: https://denizbickici.github.io/thinkgraphs/
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。
