CADENet: Condition-Adaptive Asynchronous Dual-Stream Enhancement Network for Adverse Weather Perception in Autonomous Driving
For autonomous driving perception in adverse weather, CADENet offers a real-time enhancement approach that avoids retraining and additional sensors, but the reported gains are extremely low (e.g., F1=0.023 on snow), indicating very limited practical improvement.
CADENet addresses the latency bottleneck of enhancement-then-detect pipelines for autonomous driving in adverse weather by introducing a training-free, asynchronous dual-stream system that runs detection at full frame rate without blocking. On 1327 DAWN images, it achieves a recall of 0.0103 (micro) and F1 of 0.0230 on snow, with detection sustained at ~44 FPS.
Adverse weather (rain, fog, sand, and snow) degrades camera-based object detection in autonomous vehicles. Existing enhancement-then-detect approaches stall the safety-critical perception loop, violating hard real-time requirements. Progress on this problem is also constrained by an under-recognized evaluation ceiling: ground truth annotated on degraded images cannot credit a detector that recovers objects the annotators themselves could not see, so a genuinely useful enhancement can register as a near-flat F1 gain. This paper presents CADENet (Condition-Adaptive Asynchronous Dual-stream Enhancement Network), a training-free three-thread system: Thread S (YOLOv11n) delivers detections at full frame rate with zero added latency; Thread Q applies condition-adaptive enhancement (CAPE) and fuses results via entropy-guided NMS (EG-NMS) without blocking Thread S; Thread E provides CLIP zero-shot weather classification, so new weather categories require only a new text prompt, with no labeled data and no retraining. Evaluated on 1327 DAWN images (YOLOv11m, IoU = 0.5, confidence = 0.25), CADENet achieves Recall = 0.0103 (micro), F1 = 0.0230 on snow, and F1 = 0.0038 on rain. We formalize the annotation completeness bias on DAWN-class data, so the reported F1 values are lower bounds on the true gain; recall is the annotation-gap-immune headline metric. Thread S sustains approximately 44 FPS regardless of enhancement load. No model retraining or additional sensor hardware is required.