TY - GEN
T1 - Robust Traffic Scene Segmentation Under Complex Weather Conditions
T2 - 5th International Conference on Autonomous Unmanned Systems, ICAUS 2025
AU - Hou, Yanbo
AU - Hei, Yuekun
AU - Lin, Chunmian
AU - Duan, Xuting
N1 - Publisher Copyright:
© Beijing HIWING Scientific and Technological Information Institute 2026.
PY - 2026
Y1 - 2026
N2 - This paper presents a systematic evaluation framework to assess the performance of the Segment Anything Model (SAM) in traffic scene image segmentation under diverse meteorological conditions. Addressing the critical challenge of environmental perception in autonomous driving systems, this study makes three primary contributions. First, we construct a traffic scene dataset comprising 1,350 high-resolution images (1920 × 1080) with pixel-level annotations, covering nine distinct weather-light combinations, including various conditions of rain, fog, and illumination. Second, this paper designs a novel evaluation system that incorporates adaptive preprocessing and optimized prompting strategies to thoroughly examine SAM’s zero-shot segmentation capability. The experimental results demonstrate that SAM performs remarkably well across a variety of complex scenarios, especially in daytime rainy conditions, where the mean Intersection over Union (mIoU) for vehicle segmentation reaches 82.8%. Moreover, SAM maintains robust performance in low-visibility scenarios. This study showcases SAM’s superior generalization ability through the following three aspects: outperforming baseline models across all weather conditions; effectively handling small objects in complex scenes; and possessing real-time processing capabilities.
AB - This paper presents a systematic evaluation framework to assess the performance of the Segment Anything Model (SAM) in traffic scene image segmentation under diverse meteorological conditions. Addressing the critical challenge of environmental perception in autonomous driving systems, this study makes three primary contributions. First, we construct a traffic scene dataset comprising 1,350 high-resolution images (1920 × 1080) with pixel-level annotations, covering nine distinct weather-light combinations, including various conditions of rain, fog, and illumination. Second, this paper designs a novel evaluation system that incorporates adaptive preprocessing and optimized prompting strategies to thoroughly examine SAM’s zero-shot segmentation capability. The experimental results demonstrate that SAM performs remarkably well across a variety of complex scenarios, especially in daytime rainy conditions, where the mean Intersection over Union (mIoU) for vehicle segmentation reaches 82.8%. Moreover, SAM maintains robust performance in low-visibility scenarios. This study showcases SAM’s superior generalization ability through the following three aspects: outperforming baseline models across all weather conditions; effectively handling small objects in complex scenes; and possessing real-time processing capabilities.
KW - Autonomous Driving
KW - Evaluation Framework
KW - Mean Intersection over Union (mIoU)
KW - Meteorological Conditions
KW - Segment Anything Model (SAM)
KW - Traffic Scene Image Segmentation
KW - Weather-Light Combinations
KW - Zero-Shot Segmentation
UR - https://www.scopus.com/pages/publications/105040371260
U2 - 10.1007/978-981-95-7648-7_5
DO - 10.1007/978-981-95-7648-7_5
M3 - 会议稿件
AN - SCOPUS:105040371260
SN - 9789819576470
T3 - Lecture Notes in Electrical Engineering
SP - 54
EP - 63
BT - Proceedings of 5th 2025 International Conference on Autonomous Unmanned Systems (ICAUS)
A2 - Xie, Shaorong
A2 - Niu, Yifeng
A2 - Fu, Wenxing
A2 - Qu, Yi
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 17 October 2025 through 19 October 2025
ER -