Fig. 6From: A novel Multi-Layer Attention Framework for visual description prediction using bidirectional LSTMBidirectional LSTM architectureBack to article page