File size: 1,201 Bytes
83c00db 7952898 83c00db | 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 33 34 35 36 37 38 39 | import torch
from PIL import Image
from models.base import OCRModel
from transformers import AutoModelForVision2Seq, AutoProcessor, pipeline
class TrOCROCR(OCRModel):
"""
TrOCR implementation using Hugging Face
"""
def __init__(
self,
model_name: str = "microsoft/trocr-base-handwritten",
device: str | None = None,
):
self.device = device or ("cuda" if torch.cuda.is_available() else "cpu")
self.processor = AutoProcessor.from_pretrained(model_name)
self.model = AutoModelForVision2Seq.from_pretrained(model_name)
# self.model = AutoModelForVision2Seq.from_pretrained(model_name, torch_dtype=torch.float16)
self.model.to(self.device)
self.model.eval()
@torch.no_grad()
def predict(self, image: Image.Image) -> str:
# image = preprocess(image)
pixel_values = self.processor(
images=image,
return_tensors="pt"
).pixel_values.to(self.device)
generated_ids = self.model.generate(pixel_values)
text = self.processor.batch_decode(
generated_ids,
skip_special_tokens=True
)[0]
return text.strip() |