OCR related points

1. Before ever trialed BaiD ocr, via API, seems good, could well recognize the captcha char
2. Will have a POC on HuaW ocr, see whether really self-evolve and improve the accuracy
3. main concern is for complex business documents, if not well recognize the key/fields and corresponding values, that introduce much more manual spelling check efforts.
4. wonder whether combine OCR , with neutral network deep learning with model , to have better recognition accuracy for the business document, fields and values

2021Feb23 Morning 6:43AM