Confused with ocr training about Text Detection and Text Recognition

[chocolatedesue]

I am new to ocr and happy to find this awesome works !

Acrodding to tutorials, the training and infering of the Text Recognition don't need the detection model while in the deployment both model are needed.

Since I can get the texts directly from the Text Recognition, why the detection model is needed ?

What is the relationship between two of them ?

Hope for your reply !

[chocolatedesue]

I figure it out with aistudio courses.

detection model is used to quickly get text from images

[sumitclearquote]

@chocolatedesue Hi, How does the recognition model train without detecting the text first? Could you elaborate on how you trained the text detection and text recognition models.