Abstract: Generalizing to unseen scenes presents a significant challenge to semantic segmentation models, as it aims to replicate the human adaptability. Existing deep learning models still suffer ...