Fig. 7
From: Towards unbiased skin cancer classification using deep feature fusion
SWNet architecture. Where: CONV = Convolutional layer; BN = Batch Normalization layer; Relu = Rectified Linear Unit layer; Concat = Concatenation Layer (Combines the inputs from two or more neural network layers); BNConcat = Batch Normalization Concatenation layer; Drop = Dropout layer; FC = Fully Connected Layer