RoadSceneBench：面向中层道路场景理解的轻量级基准
RoadSceneBench: A Lightweight Benchmark for Mid-Level Road Scene Understanding
为视觉语言模型在道路场景中的关系推理与结构一致性提供了轻量评测基准，并给出基于分层关系奖励传播与时序一致性的训练框架，可直接用于评估和提升 VLM 的空间与几何推理能力。
原文依据 we present RoadSceneBench, a lightweight yet information-rich benchmark designed to evaluate and advance visual reasoning in complex road environments
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论理解中层道路语义——即捕捉连接低层感知与高层规划的结构与上下文线索——对于可靠的自动驾驶和数字地图构建至关重要。然而，现有基准主要面向检测或分割等感知任务，忽视了推断道路拓扑与动态场景结构所需的推理能力。为填补这一空白，我们提出 RoadSceneBench，一个轻量却信息丰富的基准，旨在评估并推进复杂道路环境中的视觉推理。与大规模感知数据集不同，RoadSceneBench 强调关系理解与结构一致性，鼓励模型捕捉真实道路场景的底层逻辑。此外，为提升推理可靠性，我们提出分层关系奖励传播与时序一致性（HRRP-T），一种面向视觉语言模型（VLM）的训练框架，其中奖励信号在整个推理过程中自适应地促进空间连贯性与语义对齐。该范式使模型超越静态识别，迈向几何感知与时序一致的推理。大量实验表明，我们的方法在多种道路配置下均达到最优性能。因此，RoadSceneBench 为研究中层道路语义、促进结构感知的自动驾驶感知提供了紧凑而有力的基础。我们的数据集可在 https://github.com/XiyanLiu/RoadSceneBench 获取。
英文原文摘要 1568 字符
Understanding mid-level road semantics, which capture the structural and contextual cues that link low-level perception to high-level planning, is essential for reliable autonomous driving and digital map construction. However, existing benchmarks primarily target perception tasks such as detection or segmentation, overlooking the reasoning capabilities required to infer road topology and dynamic scene structure. To address this gap, we present RoadSceneBench, a lightweight yet information-rich benchmark designed to evaluate and advance visual reasoning in complex road environments. Unlike large-scale perception datasets, RoadSceneBench emphasizes relational understanding and structural consistency, encouraging models to capture the underlying logic of real-world road scenes. Furthermore, to enhance reasoning reliability, we propose Hierarchical Relational Reward Propagation with Temporal Consistency (HRRP-T), a training framework for Vision-Language Models (VLMs) in which reward signals adaptively promote spatial coherence and semantic alignment throughout the reasoning process. This paradigm enables models to move beyond static recognition toward geometry-aware and temporally consistent reasoning. Extensive experiments demonstrate that our method achieves state-of-the-art performance across diverse road configurations. RoadSceneBench thus provides a compact yet powerful foundation for studying mid-level road semantics and fostering structure-aware autonomous perception. Our dataset is available at https://github.com/XiyanLiu/RoadSceneBench.