Authors: Supei Si

Description: the result was gotten by using the cnn method Resnet101 , which was modified a little. the most common characters were Divided into 1000 categories.