[REQ] Working Hardsub OCR

[MarcoRavich]

Seems that @oliverfei finally implemented it:

https://colab.research.google.com/github/oliverfei/videocr-PaddleOCR/blob/master/videocr_PaddleOCR.ipynb

What about including it in MixLab ?

Hope that inspires !