RoadSceneBench: A Lightweight Benchmark for Mid-Level Road Scene Understanding

TL;DR

RoadSceneBench is a lightweight benchmark designed to evaluate and improve visual reasoning about road scenes, emphasizing relational understanding and structural consistency for autonomous driving applications.

Contribution

The paper introduces RoadSceneBench, a novel benchmark focusing on mid-level road scene reasoning, and proposes HRRP-T, a training framework enhancing spatial and semantic coherence in vision-language models.

Findings

Achieves state-of-the-art performance on diverse road configurations

Promotes geometry-aware and temporally consistent reasoning

Provides a compact, information-rich dataset for mid-level semantics

Abstract

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.