Table 6.
The convolutional layer parameters are denoted as “conv2 (dimension of output, number of filters)” where “conv2” denotes 2D convolution operation. “conv2-T (dimension of output, number of filters)” denotes 2D transpose convolution layer. The conv2 and conv2-T operations have ‘same' padding (i.e., output and input of the operation have the same height and width).