HRVM-UNet: Dual-path vision mamba U-Net with frequency-aware skip fusion for high-resolution remote sensing semantic segmentation

Key Points

• HRVM-UNet achieves high-resolution semantic segmentation with enhanced detail and accuracy,
• The model leverages frequency-aware skip fusion to improve information flow across layers,
• Observational analysis showcases the effectiveness of the dual-path design for capturing spatial features,
• This may enable more precise applications in remote sensing, highlighting the importance of advanced semantic segmentation methods.