In this article, we introduce DCANet, a dual-branch cross-scale feature aggregation network based on an encoder–decoder framework, incorporating visual-state-space (VSS) blocks in the encoder branch ...