Generating meaningful captions of images , through a hybrid CNN+RNN and a CNN+LSTM model