S2MFormer: Spatial-spectral Mamba-transformer complementary network for hyperspectral image classification

Key Points

• Hyperspectral image classification accuracy significantly improves with the mamba-transformer model—indicating enhanced data processing capabilities.
• The model achieves an accuracy increase to approximately 95% across diverse hyperspectral imagery datasets.
• Assessment utilizes a complementary network approach that combines spatial and spectral features for better representation.
• The findings underscore the potential of deep learning in enhancing hyperspectral image analysis for various applications.