SnowRef-Drive: A Global Instruction-Driven Traffic Video Segmentation Dataset for Adverse Weather Driving Scenarios

Engineering notes

The dataset provides a standardized benchmark for evaluating instruction grounding, temporal consistency, and segmentation robustness in safety-critical winter driving scenarios.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Adverse weather, particularly snowfall and low visibility, severely degrades vision-based traffic perception, yet existing language-driven video segmentation datasets rarely provide pixel-level annotations for real-world driving under such conditions. We present SnowRef-Drive, a large-scale global dataset for instruction-driven traffic video segmentation in adverse winter environments. It contains 19,700 short clips (118,200 annotated frames), each sampled at 2 FPS over 3 seconds (6 frames), paired with a natural language segmentation instruction and frame-wise instance masks. SnowRef-Drive spans 20 winter driving regions across North America, Europe, and East Asia, covering diverse road topologies (urban, highway, mountain/alpine, and mixed transitions) and challenging conditions including heavy snowfall, nighttime snow, fog, and low illumination. The dataset provides a standardized benchmark for evaluating instruction grounding, temporal consistency, and segmentation robustness in safety-critical winter driving scenarios.

Links and sources

• PDF from original source

Comments