Overview
The dataset consists of 332987 Chinese characters (1K classes). Train the best classifier based on cnn to outperform other participants.