A Distinguishable Pseudo-Feature Synthesis Method for Generalized Zero-Shot Learning

Figure 4

Illustration of network structures of the embedding network, the preclassifier and the classifier. In the embedding network, the dimensions of the input and the output features are marked on the left side and the right side of the FC layers, respectively. “V-D”, “S-D”, “C-D”, and “Dim” are the dimensions of visual features, the output features of the embedding network, the class number, and the middle layer dimension of the preclassifier/classifier, respectively.