Additive content and style disentanglement for domain generalized semantic segmentation of optical remote sensing images

Official URL: https://dx.doi.org/10.1109/SIU66497.2025.11111923

Supervised learning methods assume training and test data are independently and identically distributed, commonly used in land cover semantic segmentation. However, this assumption doesn't hold in real-world scenarios, as test data stems from unseen domains, leading to domain shifts that degrade model performance. Domain Generalization (DG) techniques address this by focusing on learning domain-invariant features. This paper investigates DG for land cover semantic segmentation in remote sensing optical images by progressively separating content and style features, emphasizing domain-invariant ones. Unlike state-of-the-art methods, it explores extracting only domain-invariant features. The approach is validated on the FLAIR dataset, achieving superior performance compared to existing methods.