Fig. 6.

a GoogLeNet network topology that we used for this study. b The inception module, utilized in GoogLeNet, contains six convolutional layers with different kernel sizes and a pooling layer. All resultant outputs are concatenated into a single output vector