def test_save_load_pretrained_additional_features(self):
processor = LayoutLMv2Processor(feature_extractor=self.get_feature_extractor(), tokenizer=self.get_tokenizer())
processor.save_pretrained(self.tmpdirname)
# slow tokenizer
tokenizer_add_kwargs = self.get_tokenizer(bos_token="(BOS)", eos_token="(EOS)")
feature_extractor_add_kwargs = self.get_feature_extractor(do_resize=False, size=30)
processor = LayoutLMv2Processor.from_pretrained(
self.tmpdirname, use_fast=False, bos_token="(BOS)", eos_token="(EOS)", do_resize=False, size=30
)
self.assertEqual(processor.tokenizer.get_vocab(), tokenizer_add_kwargs.get_vocab())
self.assertIsInstance(processor.tokenizer, LayoutLMv2Tokenizer)
self.assertEqual(processor.feature_extractor.to_json_string(), feature_extractor_add_kwargs.to_json_string())
self.assertIsInstance(processor.feature_extractor, LayoutLMv2FeatureExtractor)
# fast tokenizer
tokenizer_add_kwargs = self.get_rust_tokenizer(bos_token="(BOS)", eos_token="(EOS)")
feature_extractor_add_kwargs = self.get_feature_extractor(do_resize=False, size=30)
processor = LayoutLMv2Processor.from_pretrained(
self.tmpdirname, bos_token="(BOS)", eos_token="(EOS)", do_resize=False, size=30
)
self.assertEqual(processor.tokenizer.get_vocab(), tokenizer_add_kwargs.get_vocab())
self.assertIsInstance(processor.tokenizer, LayoutLMv2TokenizerFast)
self.assertEqual(processor.feature_extractor.to_json_string(), feature_extractor_add_kwargs.to_json_string())
self.assertIsInstance(processor.feature_extractor, LayoutLMv2FeatureExtractor)
def test_save_load_pretrained_default(self):
feature_extractor = self.get_feature_extractor()
tokenizers = self.get_tokenizers()
for tokenizer in tokenizers:
processor = LayoutLMv2Processor(feature_extractor=feature_extractor, tokenizer=tokenizer)
processor.save_pretrained(self.tmpdirname)
processor = LayoutLMv2Processor.from_pretrained(self.tmpdirname)
self.assertEqual(processor.tokenizer.get_vocab(), tokenizer.get_vocab())
self.assertIsInstance(processor.tokenizer, (LayoutLMv2Tokenizer, LayoutLMv2TokenizerFast))
self.assertEqual(processor.feature_extractor.to_json_string(), feature_extractor.to_json_string())
self.assertIsInstance(processor.feature_extractor, LayoutLMv2FeatureExtractor)
def test_overflowing_tokens(self):
# In the case of overflowing tokens, test that we still have 1-to-1 mapping between the images and input_ids (sequences that are too long are broken down into multiple sequences).
from datasets import load_dataset
# set up
datasets = load_dataset("nielsr/funsd")
processor = LayoutLMv2Processor.from_pretrained(
"microsoft/layoutlmv2-base-uncased", revision="no_ocr")
def preprocess_data(examples):
images = [
Image.open(path).convert("RGB")
for path in examples["image_path"]
]
words = examples["words"]
boxes = examples["bboxes"]
word_labels = examples["ner_tags"]
encoded_inputs = processor(
images,
words,
boxes=boxes,
word_labels=word_labels,
padding="max_length",
truncation=True,
return_overflowing_tokens=True,
stride=50,
return_offsets_mapping=True,
return_tensors="pt",
)
return encoded_inputs
train_data = preprocess_data(datasets["train"])
self.assertEqual(len(train_data["image"]),
len(train_data["input_ids"]))
if filename.split(".")[-1] in ["png", "jpeg", "jpg"]:
return "image"
elif filename.split(".")[-1] in ["pdf"]:
return "pdf"
else:
return "UNSUPPORTED file type"
#-------------simple_invoice_extraction----------------------------------------
#all defined outside the api to not reload exery time
import numpy as np
from transformers import LayoutLMv2Processor, LayoutLMv2ForTokenClassification
from datasets import load_dataset
from PIL import Image, ImageDraw, ImageFont
processor = LayoutLMv2Processor.from_pretrained(
"microsoft/layoutlmv2-base-uncased")
model_simple_invoice = LayoutLMv2ForTokenClassification.from_pretrained(
"Theivaprakasham/layoutlmv2-finetuned-sroie_mod")
dataset = load_dataset("darentang/generated", split="test")
labels = dataset.features['ner_tags'].feature.names
print('labels', labels)
id2label = {v: k for v, k in enumerate(labels)}
label2color = {
'b-abn': "blue",
'b-biller': "blue",
'b-biller_address': "black",
'b-biller_post_code': "green",
'b-due_date': "orange",
'b-gst': 'red',
'b-invoice_date': 'red',
