Fig. 3From: A cascading learning method with SegFormer for radiographic measurement of periodontal bone lossDiagram of the SegFormer model. SegFormer incorporates four transformer blocks, represented by the orange parts, with each capturing feature at scales of 1/4, 1/8, 1/16, and 1/32 of the original image dimensions. These features were unified to a consistent scale using an MLP layer and then processed through another MLP layer for classification. H and W indicate the input image’s height and width, respectively. C1 to C4 and C denote the feature map channel counts. Ncls specifies the class count. MLP, multilayer perceptronBack to article page