The automatic feature extraction for automatic annotation was the emphasis of this paper. And they employed a deep learning convolutional neural network to build and improve image coding and ...