CoT Block: A Transformer-Style Feature Extraction Block for Enhanced Visual Representation

This paper proposes the CoT block, which is a Transformer-style feature extraction block that not only has the advantages of self-attention mechanism but also can connect the contextual information of nearby convolution, enhancing the visual expression ability.

This paper proposes a multi-scale feature fusion module, where features at different scales contain and express different feature information. By fusing the multi-layer features of the decoding part, it can simultaneously enhance the expression of spatial geometric feature information and semantic feature information, enabling the network to accurately segment the target.