What is the best OCR (Optical character recognition). Need open source alternative to PDNod and Google [D]

PDnod translation is incredibly accurate and good. I want one that can process small screenshots of sentences that are accurate.

I tested out some like Tesseract and they just can't get it right when trying Chinese text in a video. Does anyone know a better open-source alternative to Tesseract.

character recognition chinese good google machinelearning ocr open source optical optical character recognition process recognition small tesseract text translation video