Spaces:
Runtime error
Runtime error
Commit
·
f0a21d6
1
Parent(s):
05957fd
refactor code
Browse files
app.py
CHANGED
@@ -2,6 +2,7 @@ import io
|
|
2 |
import os
|
3 |
import boto3
|
4 |
import traceback
|
|
|
5 |
|
6 |
import gradio as gr
|
7 |
from PIL import Image, ImageDraw
|
@@ -10,43 +11,37 @@ from docquery.document import load_document, ImageDocument
|
|
10 |
from docquery.ocr_reader import get_ocr_reader
|
11 |
from transformers import AutoTokenizer, AutoModelForQuestionAnswering
|
12 |
from transformers import DonutProcessor, VisionEncoderDecoderModel
|
|
|
13 |
|
14 |
# avoid ssl errors
|
15 |
import ssl
|
|
|
16 |
ssl._create_default_https_context = ssl._create_unverified_context
|
17 |
|
18 |
os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
19 |
|
|
|
20 |
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
|
|
|
|
27 |
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
"
|
33 |
-
|
34 |
-
"LiLT": "philschmid/lilt-en-funsd",
|
35 |
-
# "LiLT" : "nielsr/lilt-xlm-roberta-base"
|
36 |
-
}
|
37 |
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
# global PIPELINES
|
43 |
-
# if model in PIPELINES:
|
44 |
-
# return PIPELINES[model]
|
45 |
-
#
|
46 |
-
# device = "cuda" if torch.cuda.is_available() else "cpu"
|
47 |
-
# ret = pipeline(task=task, model=CHECKPOINTS[model], device=device)
|
48 |
-
# PIPELINES[model] = ret
|
49 |
-
# return ret
|
50 |
|
51 |
|
52 |
def image_to_byte_array(image: Image) -> bytes:
|
@@ -56,25 +51,25 @@ def image_to_byte_array(image: Image) -> bytes:
|
|
56 |
return image_as_byte_array
|
57 |
|
58 |
|
59 |
-
def
|
60 |
image_as_byte_base64 = image_to_byte_array(image=document.b)
|
61 |
-
response = boto3.client(
|
62 |
Document={
|
63 |
-
|
64 |
},
|
65 |
FeatureTypes=[
|
66 |
-
|
67 |
],
|
68 |
QueriesConfig={
|
69 |
-
|
70 |
{
|
71 |
-
|
72 |
-
|
73 |
-
|
74 |
-
]
|
75 |
},
|
76 |
]
|
77 |
-
}
|
78 |
)
|
79 |
for element in response["Blocks"]:
|
80 |
if element["BlockType"] == "QUERY_RESULT":
|
@@ -87,75 +82,60 @@ def run_textract_query(question, document):
|
|
87 |
Exception("No QUERY_RESULT found in the response from Textract.")
|
88 |
|
89 |
|
90 |
-
def
|
91 |
-
|
92 |
-
|
93 |
-
nlp = pipeline(
|
94 |
-
"document-question-answering",
|
95 |
-
model="impira/layoutlm-document-qa",
|
96 |
-
)
|
97 |
-
|
98 |
-
result = nlp(document.context["image"][0][0], question)[0]
|
99 |
# [{'score': 0.9999411106109619, 'answer': 'LETTER OF CREDIT', 'start': 106, 'end': 108}]
|
100 |
return {
|
101 |
"score": result["score"],
|
102 |
"answer": result["answer"],
|
103 |
"word_ids": [result["start"], result["end"]],
|
104 |
-
"page": 0
|
105 |
}
|
106 |
|
107 |
|
108 |
-
def
|
109 |
-
|
110 |
# use this model + tokenizer
|
111 |
-
lilt_tokenizer = AutoTokenizer.from_pretrained("SCUT-DLVCLab/lilt-infoxlm-base")
|
112 |
-
model = AutoModelForQuestionAnswering.from_pretrained("nielsr/lilt-xlm-roberta-base")
|
113 |
-
|
114 |
processed_document = document.context["image"][0][1]
|
115 |
words = [x[0] for x in processed_document]
|
116 |
boxes = [x[1] for x in processed_document]
|
117 |
|
118 |
-
encoding = lilt_tokenizer(
|
119 |
-
|
120 |
-
|
|
|
|
|
|
|
|
|
|
|
121 |
|
122 |
answer_start_index = outputs.start_logits.argmax()
|
123 |
answer_end_index = outputs.end_logits.argmax()
|
124 |
|
125 |
-
predict_answer_tokens = encoding.input_ids[
|
126 |
-
|
|
|
|
|
|
|
|
|
127 |
return {
|
128 |
-
|
129 |
-
|
130 |
-
|
131 |
-
|
132 |
|
133 |
|
134 |
def run_donut(question, document):
|
135 |
-
|
136 |
-
# nlp = pipeline(
|
137 |
-
# "document-question-answering",
|
138 |
-
# model="naver-clova-ix/donut-base-finetuned-docvqa",
|
139 |
-
# )
|
140 |
-
#
|
141 |
-
# result = nlp(document.context["image"][0][0], question)[0]
|
142 |
-
# # [{'score': 0.9999411106109619, 'answer': 'LETTER OF CREDIT', 'start': 106, 'end': 108}]
|
143 |
-
# return {
|
144 |
-
# "score": result["score"],
|
145 |
-
# "answer": result["answer"],
|
146 |
-
# "word_ids": [result["start"], result["end"]],
|
147 |
-
# "page": 0
|
148 |
-
# }
|
149 |
-
|
150 |
-
donut_processor = DonutProcessor.from_pretrained("naver-clova-ix/donut-base-finetuned-docvqa")
|
151 |
-
donut_model = VisionEncoderDecoderModel.from_pretrained("naver-clova-ix/donut-base-finetuned-docvqa")
|
152 |
# prepare encoder inputs
|
153 |
-
pixel_values = donut_processor(
|
|
|
|
|
154 |
|
155 |
# prepare decoder inputs
|
156 |
task_prompt = "<s_docvqa><s_question>{user_input}</s_question><s_answer>"
|
157 |
prompt = task_prompt.replace("{user_input}", question)
|
158 |
-
decoder_input_ids = donut_processor.tokenizer(
|
|
|
|
|
159 |
|
160 |
# generate answer
|
161 |
outputs = donut_model.generate(
|
@@ -170,11 +150,13 @@ def run_donut(question, document):
|
|
170 |
bad_words_ids=[[donut_processor.tokenizer.unk_token_id]],
|
171 |
return_dict_in_generate=True,
|
172 |
)
|
173 |
-
import re
|
174 |
-
# postprocess
|
175 |
sequence = donut_processor.batch_decode(outputs.sequences)[0]
|
176 |
-
sequence = sequence.replace(donut_processor.tokenizer.eos_token, "").replace(
|
177 |
-
|
|
|
|
|
|
|
|
|
178 |
|
179 |
result = donut_processor.token2json(sequence)
|
180 |
return {
|
@@ -184,26 +166,6 @@ def run_donut(question, document):
|
|
184 |
}
|
185 |
|
186 |
|
187 |
-
def run_pipeline(model, question, document, top_k):
|
188 |
-
""" Run pipeline selected by the user.
|
189 |
-
:return: expect an object like
|
190 |
-
[{'score': 0.251716673374176, 'answer': 'CREDIT', 'word_ids': [38], 'page': 0},
|
191 |
-
{'score': 0.15292450785636902, 'answer': 'LETTER OF CREDIT', 'word_ids': [37, 38], 'page': 0},
|
192 |
-
{'score': 0.009600160643458366, 'answer': 'Payment Tens LETTER OF CREDIT', 'word_ids': [36, 37, 38], 'page': 0}]
|
193 |
-
"""
|
194 |
-
if model == "Textract Query":
|
195 |
-
return run_textract_query(question, document)
|
196 |
-
elif model == "LiLT":
|
197 |
-
return run_lilt_model(question, document)
|
198 |
-
elif model == "LayoutLM FineTuned":
|
199 |
-
return run_layoutlm_finetuned(question=question, document=document)
|
200 |
-
elif model == "Donut":
|
201 |
-
return run_donut(question=question, document=document)
|
202 |
-
else:
|
203 |
-
return {"answer": "model not found", "score": "n/a"}
|
204 |
-
|
205 |
-
|
206 |
-
|
207 |
def process_path(path):
|
208 |
error = None
|
209 |
if path:
|
@@ -230,6 +192,7 @@ def process_path(path):
|
|
230 |
None,
|
231 |
)
|
232 |
|
|
|
233 |
def process_upload(file):
|
234 |
if file:
|
235 |
return process_path(file.name)
|
@@ -268,11 +231,19 @@ def normalize_bbox(box, width, height, padding=0.005):
|
|
268 |
return [min_x * width, min_y * height, max_x * width, max_y * height]
|
269 |
|
270 |
|
271 |
-
|
272 |
-
|
273 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
274 |
text_value = prediction["answer"]
|
275 |
if "word_ids" in prediction:
|
|
|
276 |
image = pages[prediction["page"]]
|
277 |
draw = ImageDraw.Draw(image, "RGBA")
|
278 |
word_boxes = lift_word_boxes(document, prediction["page"])
|
@@ -397,7 +368,6 @@ gradio-app h2, .gradio-app h2 {
|
|
397 |
"""
|
398 |
|
399 |
examples = [
|
400 |
-
|
401 |
[
|
402 |
"scenario-1.png",
|
403 |
"What is the final consignee?",
|
@@ -416,7 +386,7 @@ examples = [
|
|
416 |
],
|
417 |
[
|
418 |
"scenario-4.png",
|
419 |
-
|
420 |
],
|
421 |
[
|
422 |
"scenario-5.png",
|
@@ -458,9 +428,7 @@ examples = [
|
|
458 |
|
459 |
with gr.Blocks(css=CSS) as demo:
|
460 |
gr.Markdown("# Document Query Engine")
|
461 |
-
gr.Markdown(
|
462 |
-
"Original version comes from DocQuery [here](https://huggingface.co/spaces/impira/docquery) (created by [Impira](https://impira.com?utm_source=huggingface&utm_medium=referral&utm_campaign=docquery_space))"
|
463 |
-
)
|
464 |
|
465 |
document = gr.Variable()
|
466 |
example_question = gr.Textbox(visible=False)
|
@@ -489,8 +457,8 @@ with gr.Blocks(css=CSS) as demo:
|
|
489 |
max_lines=1,
|
490 |
)
|
491 |
model = gr.Radio(
|
492 |
-
choices=list(
|
493 |
-
value=list(
|
494 |
label="Model",
|
495 |
)
|
496 |
|
|
|
2 |
import os
|
3 |
import boto3
|
4 |
import traceback
|
5 |
+
import re
|
6 |
|
7 |
import gradio as gr
|
8 |
from PIL import Image, ImageDraw
|
|
|
11 |
from docquery.ocr_reader import get_ocr_reader
|
12 |
from transformers import AutoTokenizer, AutoModelForQuestionAnswering
|
13 |
from transformers import DonutProcessor, VisionEncoderDecoderModel
|
14 |
+
from transformers import pipeline
|
15 |
|
16 |
# avoid ssl errors
|
17 |
import ssl
|
18 |
+
|
19 |
ssl._create_default_https_context = ssl._create_unverified_context
|
20 |
|
21 |
os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
22 |
|
23 |
+
# Init models
|
24 |
|
25 |
+
layoutlm_pipeline = pipeline(
|
26 |
+
"document-question-answering",
|
27 |
+
model="impira/layoutlm-document-qa",
|
28 |
+
)
|
29 |
+
lilt_tokenizer = AutoTokenizer.from_pretrained("SCUT-DLVCLab/lilt-infoxlm-base")
|
30 |
+
lilt_model = AutoModelForQuestionAnswering.from_pretrained(
|
31 |
+
"nielsr/lilt-xlm-roberta-base"
|
32 |
+
)
|
33 |
|
34 |
+
donut_processor = DonutProcessor.from_pretrained(
|
35 |
+
"naver-clova-ix/donut-base-finetuned-docvqa"
|
36 |
+
)
|
37 |
+
donut_model = VisionEncoderDecoderModel.from_pretrained(
|
38 |
+
"naver-clova-ix/donut-base-finetuned-docvqa"
|
39 |
+
)
|
|
|
|
|
|
|
40 |
|
41 |
+
TEXTRACT = "Textract Query"
|
42 |
+
LAYOUTLM = "LayoutLM"
|
43 |
+
DONUT = "Donut"
|
44 |
+
LILT = "LiLT"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
|
46 |
|
47 |
def image_to_byte_array(image: Image) -> bytes:
|
|
|
51 |
return image_as_byte_array
|
52 |
|
53 |
|
54 |
+
def run_textract(question, document):
|
55 |
image_as_byte_base64 = image_to_byte_array(image=document.b)
|
56 |
+
response = boto3.client("textract").analyze_document(
|
57 |
Document={
|
58 |
+
"Bytes": image_as_byte_base64,
|
59 |
},
|
60 |
FeatureTypes=[
|
61 |
+
"QUERIES",
|
62 |
],
|
63 |
QueriesConfig={
|
64 |
+
"Queries": [
|
65 |
{
|
66 |
+
"Text": question,
|
67 |
+
"Pages": [
|
68 |
+
"*",
|
69 |
+
],
|
70 |
},
|
71 |
]
|
72 |
+
},
|
73 |
)
|
74 |
for element in response["Blocks"]:
|
75 |
if element["BlockType"] == "QUERY_RESULT":
|
|
|
82 |
Exception("No QUERY_RESULT found in the response from Textract.")
|
83 |
|
84 |
|
85 |
+
def run_layoutlm(question, document):
|
86 |
+
result = layoutlm_pipeline(document.context["image"][0][0], question)[0]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
87 |
# [{'score': 0.9999411106109619, 'answer': 'LETTER OF CREDIT', 'start': 106, 'end': 108}]
|
88 |
return {
|
89 |
"score": result["score"],
|
90 |
"answer": result["answer"],
|
91 |
"word_ids": [result["start"], result["end"]],
|
92 |
+
"page": 0,
|
93 |
}
|
94 |
|
95 |
|
96 |
+
def run_lilt(question, document):
|
|
|
97 |
# use this model + tokenizer
|
|
|
|
|
|
|
98 |
processed_document = document.context["image"][0][1]
|
99 |
words = [x[0] for x in processed_document]
|
100 |
boxes = [x[1] for x in processed_document]
|
101 |
|
102 |
+
encoding = lilt_tokenizer(
|
103 |
+
text=question,
|
104 |
+
text_pair=words,
|
105 |
+
boxes=boxes,
|
106 |
+
add_special_tokens=True,
|
107 |
+
return_tensors="pt",
|
108 |
+
)
|
109 |
+
outputs = lilt_model(**encoding)
|
110 |
|
111 |
answer_start_index = outputs.start_logits.argmax()
|
112 |
answer_end_index = outputs.end_logits.argmax()
|
113 |
|
114 |
+
predict_answer_tokens = encoding.input_ids[
|
115 |
+
0, answer_start_index: answer_end_index + 1
|
116 |
+
]
|
117 |
+
predict_answer = lilt_tokenizer.decode(
|
118 |
+
predict_answer_tokens, skip_special_tokens=True
|
119 |
+
)
|
120 |
return {
|
121 |
+
"score": "n/a",
|
122 |
+
"answer": predict_answer,
|
123 |
+
# "word_ids": element
|
124 |
+
}
|
125 |
|
126 |
|
127 |
def run_donut(question, document):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
128 |
# prepare encoder inputs
|
129 |
+
pixel_values = donut_processor(
|
130 |
+
document.context["image"][0][0], return_tensors="pt"
|
131 |
+
).pixel_values
|
132 |
|
133 |
# prepare decoder inputs
|
134 |
task_prompt = "<s_docvqa><s_question>{user_input}</s_question><s_answer>"
|
135 |
prompt = task_prompt.replace("{user_input}", question)
|
136 |
+
decoder_input_ids = donut_processor.tokenizer(
|
137 |
+
prompt, add_special_tokens=False, return_tensors="pt"
|
138 |
+
).input_ids
|
139 |
|
140 |
# generate answer
|
141 |
outputs = donut_model.generate(
|
|
|
150 |
bad_words_ids=[[donut_processor.tokenizer.unk_token_id]],
|
151 |
return_dict_in_generate=True,
|
152 |
)
|
|
|
|
|
153 |
sequence = donut_processor.batch_decode(outputs.sequences)[0]
|
154 |
+
sequence = sequence.replace(donut_processor.tokenizer.eos_token, "").replace(
|
155 |
+
donut_processor.tokenizer.pad_token, ""
|
156 |
+
)
|
157 |
+
sequence = re.sub(
|
158 |
+
r"<.*?>", "", sequence, count=1
|
159 |
+
).strip() # remove first task start token
|
160 |
|
161 |
result = donut_processor.token2json(sequence)
|
162 |
return {
|
|
|
166 |
}
|
167 |
|
168 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
169 |
def process_path(path):
|
170 |
error = None
|
171 |
if path:
|
|
|
192 |
None,
|
193 |
)
|
194 |
|
195 |
+
|
196 |
def process_upload(file):
|
197 |
if file:
|
198 |
return process_path(file.name)
|
|
|
231 |
return [min_x * width, min_y * height, max_x * width, max_y * height]
|
232 |
|
233 |
|
234 |
+
MODELS = {
|
235 |
+
TEXTRACT: run_textract,
|
236 |
+
LAYOUTLM: run_layoutlm,
|
237 |
+
DONUT: run_donut,
|
238 |
+
LILT: run_lilt,
|
239 |
+
}
|
240 |
+
|
241 |
+
|
242 |
+
def process_question(question, document, model=list(MODELS.keys())[0]):
|
243 |
+
prediction = MODELS[model](question=question, document=document)
|
244 |
text_value = prediction["answer"]
|
245 |
if "word_ids" in prediction:
|
246 |
+
pages = [x.copy().convert("RGB") for x in document.preview]
|
247 |
image = pages[prediction["page"]]
|
248 |
draw = ImageDraw.Draw(image, "RGBA")
|
249 |
word_boxes = lift_word_boxes(document, prediction["page"])
|
|
|
368 |
"""
|
369 |
|
370 |
examples = [
|
|
|
371 |
[
|
372 |
"scenario-1.png",
|
373 |
"What is the final consignee?",
|
|
|
386 |
],
|
387 |
[
|
388 |
"scenario-4.png",
|
389 |
+
"What is the color?",
|
390 |
],
|
391 |
[
|
392 |
"scenario-5.png",
|
|
|
428 |
|
429 |
with gr.Blocks(css=CSS) as demo:
|
430 |
gr.Markdown("# Document Query Engine")
|
431 |
+
gr.Markdown("### Compare performance of different document layout models. If you have any suggestions [contact me](https://www.linkedin.com/in/vincent-claes-0b346337/)")
|
|
|
|
|
432 |
|
433 |
document = gr.Variable()
|
434 |
example_question = gr.Textbox(visible=False)
|
|
|
457 |
max_lines=1,
|
458 |
)
|
459 |
model = gr.Radio(
|
460 |
+
choices=list(MODELS.keys()),
|
461 |
+
value=list(MODELS.keys())[0],
|
462 |
label="Model",
|
463 |
)
|
464 |
|