The progress was used to generate synthetic dataset for Chinese OCR. - View it on GitHub