Attention UNet: Enhancing Image Feature Extraction with Spatial Attention

The UNet model uses attention gate modules to highlight important and relevant features in images while suppressing irrelevant ones. In this model, hard attention is replaced by soft attention, and the attention module is integrated into the skip connections and upsampling modules of the UNet network to implement spatial attention mechanisms.