Spaces:
Runtime error
Runtime error
File size: 3,196 Bytes
35b22df |
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 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 |
"""Image parser.
Contains parsers for image files.
"""
import re
from pathlib import Path
from typing import Dict
from gpt_index.readers.file.base_parser import BaseParser
class ImageParser(BaseParser):
"""Image parser.
Extract text from images using DONUT.
"""
def _init_parser(self) -> Dict:
"""Init parser."""
try:
import torch # noqa: F401
except ImportError:
raise ImportError(
"install pytorch to use the model: " "`pip install torch`"
)
try:
from transformers import DonutProcessor, VisionEncoderDecoderModel
except ImportError:
raise ImportError(
"transformers is required for using DONUT model: "
"`pip install transformers`"
)
try:
import sentencepiece # noqa: F401
except ImportError:
raise ImportError(
"sentencepiece is required for using DONUT model: "
"`pip install sentencepiece`"
)
try:
from PIL import Image # noqa: F401
except ImportError:
raise ImportError(
"PIL is required to read image files: " "`pip install Pillow`"
)
processor = DonutProcessor.from_pretrained(
"naver-clova-ix/donut-base-finetuned-cord-v2"
)
model = VisionEncoderDecoderModel.from_pretrained(
"naver-clova-ix/donut-base-finetuned-cord-v2"
)
return {"processor": processor, "model": model}
def parse_file(self, file: Path, errors: str = "ignore") -> str:
"""Parse file."""
import torch
from PIL import Image
model = self.parser_config["model"]
processor = self.parser_config["processor"]
device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)
# load document image
image = Image.open(file)
if image.mode != "RGB":
image = image.convert("RGB")
# prepare decoder inputs
task_prompt = "<s_cord-v2>"
decoder_input_ids = processor.tokenizer(
task_prompt, add_special_tokens=False, return_tensors="pt"
).input_ids
pixel_values = processor(image, return_tensors="pt").pixel_values
outputs = model.generate(
pixel_values.to(device),
decoder_input_ids=decoder_input_ids.to(device),
max_length=model.decoder.config.max_position_embeddings,
early_stopping=True,
pad_token_id=processor.tokenizer.pad_token_id,
eos_token_id=processor.tokenizer.eos_token_id,
use_cache=True,
num_beams=3,
bad_words_ids=[[processor.tokenizer.unk_token_id]],
return_dict_in_generate=True,
)
sequence = processor.batch_decode(outputs.sequences)[0]
sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(
processor.tokenizer.pad_token, ""
)
# remove first task start token
sequence = re.sub(r"<.*?>", "", sequence, count=1).strip()
return sequence
|