Application of neural network in map annotation recognition

To further expand the application field of deep learning and explore the application of neural network in cartography, this paper constructed a map annotation category and annotation character recognition system based on neural network. This paper used convolution neural network and dense convolution network models, based on TensorFlow and MXNet frameworks, studied the map annotation data set constructed by obtaining pictures from publicly published atlases from category recognition and character recognition of map annotation. The results showed that in annotation classification, the classification accuracy of convolutional neural network with convolutional block attention module was 99.08%. In annotation character recognition, the recognition accuracy of dense convolution network model was 95.36%. The experimental results show that the neural network can well realize map annotation category and character recognition. The research results provide a new idea for further improving the application of neural network model in cartography.