A domain‐adaptive method with cycle perceptual consistency adversarial networks for vehicle target detection in foggy weather

Foggy weather can cause such problems as blurred image information and the loss of details, which may pose great challenges to road traffic target detection based on images videos. In this study, we propose a domain-adaptive vehicle method implement domain adaptation for real foggy scene. We firstly constructed highway dataset with (HVFD), contains normal provides complete data support computer vision. Secondly, by improving CycleGAN designed an improved generative confrontation network (CPGAN), realised style transfer between images. Finally, formulated YOLOv4 framework according pre-trained fog model. The experimental results show that put forward effectively improve performance reduce work manually labelling large number tags, has strong generalisation ability vision-based applications in low-visibility weather.