HG-Lane：在恶劣天气和光照条件下高保真生成车道场景，无需重新标注

Lane detection is a critical task in autonomous driving, ensuring the safe operation of vehicles. However, existing datasets like CULane and TuSimple have limited data under extreme weather conditions such as rain, snow, and fog. Consequently, detection models trained on these datasets often become unreliable in such environments, leading to serious safety-critical failures on the road. To address this issue, HG-Lane is proposed, a high-fidelity generation framework for lane scenes under adverse weather and lighting conditions without the need for re-annotation. Based on this framework, a benchmark is constructed that includes adverse weather and lighting scenarios, comprising 30,000 images. Experimental results demonstrate the significant effectiveness of this framework in enhancing the robustness of lane detection models.