Cross-Domain Outdoor Scene UnderstandingUsing PC-GAN and TL-SZKSE-DenseNet inAdverse Weather Conditions
A dual framework bringing together a Panoptic Cycle-Generative Adversarial Network (PC-GAN) with a Transfer Learning-based Squeeze-and-ZeroKnowledge Spatial Excitation DenseNet (TL-SZKSE-DenseNet) achieves realistic crossweather image translation with semantically aligned representations in clear and degraded scenes.