SADWA: Fine-Grained Weather Awareness with Vision-Language Models for Seamless Autonomous Driving in Real Time

Jinwoo Kim, Hayeon O, Youngmin Oh, Kyounghwan An, Donghwan Lee; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, 2025, pp. 832-841

Autonomous driving systems require adaptive driving strategies to ensure safe and efficient operation under diverse weather and road conditions. Traditional weather classification models often categorize images into broad labels such as clear, rainy, or snowy, lacking the finer granularity required for real-world driving scenarios. In this work, we fine-tune a vision-language model on a dataset that captures nuanced weather conditions, considering factors such as road surface state, precipitation intensity, visibility obstructions, and time of day. Our model leverages textual descriptions of driving environments to enhance classification accuracy while maintaining real-time inference capabilities. Experimental results demonstrate that our approach effectively distinguishes complex weather conditions, adapting to real-time environmental changes. Furthermore, we show that our fine-tuned model can accurately classify a wide range of fine-grained weather conditions while maintaining computational efficiency. This work provides a practical and scalable solution for weather-aware autonomous driving, contributing to safer and more reliable self-driving perception systems.