search

devmaxxing / videocr-PaddleOCR

Extract hardcoded subtitles from videos using machine learning
MIT License
124 stars 16 forks source link

Regarding to lang #2

Closed Soebb closed 2 years ago

Soebb commented 2 years ago

Hi and thanks for this awesome repo, Recently I got this error for 'fa' (Arabic related lang) :

AssertionError: param lang must in dict_keys(['ch', 'en', 'french', 'german', 'korean', 'japan']), but got fa

It seems the default model doesn't work for 'fa' and i need to download and specify model myself.

devmaxxing commented 2 years ago

Hello, just checking if this issue is actually resolved?

With the latest PaddleOCR release, when I tried specifying lang as 'fa', the program automatically downloaded the default arabic model and seems to work fine. They do also appear to have an 'fa' specific model listed here: https://github.com/PaddlePaddle/PaddleOCR/issues/1048