Release Note

• Release PP-OCRv3: With comparable speed, the effect of Chinese scene is further improved by 5% compared with PP-OCRv2, the effect of English scene is improved by 11%, and the average recognition accuracy of 80 language multilingual models is improved by more than 5%.
• Release PPOCRLabelv2: Add the annotation function for table recognition task, key information extraction task and irregular text image.
• Release interactive e-book "Dive into OCR", covers the cutting-edge theory and code practice of OCR full stack technology.