Research on Semantic Segmentation Methods for High-Resolution Images by Fusing Transformer Structures

High-resolution image segmentation technology plays a key role in intelligent vision systems, but traditional methods are difficult to balance the computational overhead and segmentation accuracy. In this study, we propose a high-resolution semantic segmentation network with fused transformer structure, design a multi-scale transformer encoder to capture the global context, optimise an efficient attention mechanism to reduce the computational complexity, and implement feature fusion and decoder improvement to enhance the boundary accuracy. Experiments show that the method significantly reduces memory consumption while maintaining high accuracy, providing an effective solution for real-time high-resolution semantic segmentation.