Caption Generation Based on Emotions Using CSPDenseNet and BiLSTM with Self-Attention

Figure 2

CSPDenseNet architecture with partial dense blocks and partial transition layers.