Fig. 6
From: Lightweight Unet with depthwise separable convolution for skin lesion segmentation
Channel and spatial attention weights (scaled by α/β) are applied to the input feature map for enhanced feature representation.
Fig. 6 | Scientific Reports
