The Synthetic Chinese String dataset is a Chinese character recognition dataset that contains more than 3.6 million training images and 5,824 characters. The image scenes are relatively simple, with black characters on a white background.