Figure 3
From: DLA-Net: dual lesion attention network for classification of pneumoconiosis using chest X-ray images
Architecture of the dual lesion attention (DLA) module that consists of two main submodules: channel attention (CA) and spatial attention (SA). Given the input feature map F, the CA module computes the corresponding channel-wise attention map \((F_c)\) and the SA module computes the final spatial-wise attention map \((F_{cs})\). \(F_{sq} (.):\) squeeze operation, \(F_{ex} (.):\) excitation operation, \(F_{scale} (.):\) scaling operation.
