reviseIn addition in order to integrate effective feature information from different modules and tasks an information interaction layer IIL is designed to mine the interactive information between two

Furthermore, to effectively incorporate feature information from various modules and tasks, we have introduced an information interaction layer (IIL) that aims to extract the interactive information between two subnetworks, namely defect edge detection and region segmentation. The diagram illustrating the architecture of the IIL can be seen in figure 5. For instance, when considering the fusion of edge features, we initially employ the Sigmoid function to produce weights that signify the significance of each position within the edge feature mapping