OcrModel

A loaded OCR model.

recognize

recognize(self, image: 'clika_runtime.Tensor', *, max_new_tokens: 'int' = 0) -> 'str'

The text in an [H, W, 3] image.

recognize_pages

recognize_pages(self, pages: 'Sequence[clika_runtime.Tensor]', *, max_new_tokens: 'int' = 0) -> 'str'

The text of one document, its [H, W, 3] pages in order. A model that reads one image per request refuses more than one page.