File size: 1,246 Bytes
187408a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 | from transformers import ProcessorMixin
from nav2tex.image_processing_latex_ocr import Nav2TexImageProcessor
from nav2tex.tokenization_latex_ocr import LaTeXTokenizer
class Nav2TexProcessor(ProcessorMixin):
attributes = ["image_processor", "tokenizer"]
image_processor_class = "AutoImageProcessor"
tokenizer_class = "AutoTokenizer"
def __init__(self, image_processor, tokenizer):
super().__init__(image_processor, tokenizer)
def __call__(self, images=None, text=None, return_tensors=None, **kwargs):
if images is None and text is None:
raise ValueError("You must specify either images or text.")
output = {}
if images is not None:
image_inputs = self.image_processor(images, return_tensors=return_tensors, **kwargs)
output.update(image_inputs)
if text is not None:
text_inputs = self.tokenizer(text, return_tensors=return_tensors, **kwargs)
output.update(text_inputs)
return output
def batch_decode(self, *args, **kwargs):
return self.tokenizer.batch_decode(*args, **kwargs)
def decode(self, *args, **kwargs):
return self.tokenizer.decode(*args, **kwargs) |