SAM-CTMapper: Utilizing segment anything model and scale-aware mixed CNN-Transformer facilitates coastal wetland hyperspectral image classification

Jiaqi Zou, Wei He, Haifeng Wang, Hongyan Zhang

Abstract: Highlights•We propose a wetland hyperspectral image (HSI) classification framework.•A segment anything model-based labeling strategy is devised to construct datasets.•A scale-aware mixed CNN-Transformer classification method (CTMapper) is designed.•CTMapper achieves better classification results than state-of-the art methods.