Fig. 5From: A novel Multi-Layer Attention Framework for visual description prediction using bidirectional LSTMLSTM architectural previewBack to article page