Spaces:
Sleeping
Sleeping
File size: 12,560 Bytes
0fcd216 9da9cd5 f839a97 7ca4a38 0fcd216 f839a97 0fcd216 24d5b1d 7ca4a38 0fcd216 9da9cd5 24d5b1d 9da9cd5 0fcd216 f839a97 dc9a124 8a6de15 f839a97 9da9cd5 c0d31bb 9da9cd5 c0d31bb 9da9cd5 7ca4a38 0fcd216 38e9395 9da9cd5 38e9395 9da9cd5 0fcd216 9da9cd5 0fcd216 f839a97 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 |
# import re
# import gradio as gr
# import torch
# from transformers import DonutProcessor, VisionEncoderDecoderModel
# processor = DonutProcessor.from_pretrained("pacman2223/univ-docu-model-v3")
# model = VisionEncoderDecoderModel.from_pretrained("pacman2223/univ-docu-model-v3")
# device = "cuda" if torch.cuda.is_available() else "cpu"
# model.to(device)
# def process_document(image, question):
# # prepare encoder inputs
# pixel_values = processor(image, return_tensors="pt").pixel_values
# # prepare decoder inputs
# task_prompt = "{user_input}"
# prompt = task_prompt.replace("{user_input}", question)
# decoder_input_ids = processor.tokenizer(prompt, add_special_tokens=False, return_tensors="pt").input_ids
# # generate answer
# 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=1,
# bad_words_ids=[[processor.tokenizer.unk_token_id]],
# return_dict_in_generate=True,
# )
# # postprocess
# sequence = processor.batch_decode(outputs.sequences)[0]
# sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
# sequence = re.sub(r"<.*?>", "", sequence, count=1).strip() # remove first task start token
# return processor.token2json(sequence)
# description = "Gradio Demo for Model-V3, an instance of `VisionEncoderDecoderModel` fine-tuned on DocVQA (document visual question answering). To use it, simply upload your image and type a question and click 'submit', or click one of the examples to load them. Read more at the links below."
# article = "<p style='text-align: center'>Model-V3</p>"
# demo = gr.Interface(
# fn=process_document,
# inputs=["image", "text"],
# outputs="json",
# title="Demo: Model-V3 for Document Analysis",
# description=description,
# article=article,
# examples=[["example_1.png", "What is the title shown?"], ["example_2.png", "When is mid semester exams?"]],
# cache_examples=False)
# demo.queue(max_size=5)
# demo.launch()
# import re
# import gradio as gr
# import torch
# from transformers import DonutProcessor, VisionEncoderDecoderModel
# import fitz # PyMuPDF
# from PIL import Image
# import io
# processor = DonutProcessor.from_pretrained("pacman2223/univ-docu-model-v3")
# model = VisionEncoderDecoderModel.from_pretrained("pacman2223/univ-docu-model-v3")
# device = "cuda" if torch.cuda.is_available() else "cpu"
# model.to(device)
# def pdf_to_images(pdf_file):
# if pdf_file is None:
# return None
# pdf_path = pdf_file.name # Get the file path
# images = []
# try:
# doc = fitz.open(pdf_path)
# for page in doc:
# pix = page.get_pixmap()
# img = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
# images.append(img)
# return images
# except Exception as e:
# print(f"Error converting PDF: {e}")
# return None
# def process_document(pdf_file, page_number, question):
# if pdf_file is None:
# return "Please upload a PDF file."
# images = pdf_to_images(pdf_file)
# if images is None:
# return "Failed to process the PDF file."
# if page_number < 1 or page_number > len(images):
# return f"Invalid page number. The PDF has {len(images)} pages."
# image = images[page_number - 1]
# # prepare encoder inputs
# pixel_values = processor(image, return_tensors="pt").pixel_values
# # prepare decoder inputs
# task_prompt = "{user_input}"
# prompt = task_prompt.replace("{user_input}", question)
# decoder_input_ids = processor.tokenizer(prompt, add_special_tokens=False, return_tensors="pt").input_ids
# # generate answer
# 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=1,
# bad_words_ids=[[processor.tokenizer.unk_token_id]],
# return_dict_in_generate=True,
# )
# # postprocess
# sequence = processor.batch_decode(outputs.sequences)[0]
# sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
# sequence = re.sub(r"<.*?>", "", sequence, count=1).strip() # remove first task start token
# return processor.token2json(sequence)
# def update_page_preview(pdf_file, page_number):
# if pdf_file is None:
# return None
# images = pdf_to_images(pdf_file)
# if images is None or page_number < 1 or page_number > len(images):
# return None
# return images[page_number - 1]
# # def update_page_slider(pdf_file):
# # if pdf_file is None:
# # return gr.Slider(minimum=1, maximum=1, value=1, step=1, label="Page Number")
# # images = pdf_to_images(pdf_file)
# # if images is None:
# # return gr.Slider(minimum=1, maximum=1, value=1, step=1, label="Page Number")
# # return gr.Slider(minimum=1, maximum=len(images), value=1, step=1, label="Page Number")
# description = "Gradio Demo for Model-V3, an instance of `VisionEncoderDecoderModel` fine-tuned on DocVQA (document visual question answering). To use it, upload a PDF file, select a page number, type a question, and click 'submit'."
# article = "<p style='text-align: center'>Model-V3</p>"
# with gr.Blocks() as demo:
# gr.Markdown("# Demo: Model-V3 for Document Analysis")
# gr.Markdown(description)
# with gr.Row():
# with gr.Column(scale=1):
# pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
# page_slider = gr.Slider(minimum=1, maximum=1, value=1, step=1, label="Page Number")
# with gr.Column(scale=2):
# page_preview = gr.Image(label="Page Preview")
# question_input = gr.Textbox(label="Question")
# submit_button = gr.Button("Submit")
# output = gr.JSON(label="Output")
# def update_interface(pdf_file):
# if pdf_file is None:
# return gr.Slider(minimum=1, maximum=1, value=1, step=1, label="Page Number"), None
# images = pdf_to_images(pdf_file)
# if images is None:
# return gr.Slider(minimum=1, maximum=1, value=1, step=1, label="Page Number"), None
# return (
# gr.Slider(minimum=1, maximum=len(images), value=1, step=1, label="Page Number"),
# images[0] # Show the first page by default
# )
# pdf_input.change(update_interface, inputs=[pdf_input], outputs=[page_slider, page_preview])
# page_slider.change(update_page_preview, inputs=[pdf_input, page_slider], outputs=[page_preview])
# submit_button.click(process_document, inputs=[pdf_input, page_slider, question_input], outputs=[output])
# demo.launch()
import re
import gradio as gr
import torch
from transformers import DonutProcessor, VisionEncoderDecoderModel
import fitz # PyMuPDF
from PIL import Image
import io
processor = DonutProcessor.from_pretrained("pacman2223/univ-docu-model-v3")
model = VisionEncoderDecoderModel.from_pretrained("pacman2223/univ-docu-model-v3")
device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)
def pdf_to_images(pdf_file):
if pdf_file is None:
return None
pdf_path = pdf_file.name # Get the file path
images = []
try:
doc = fitz.open(pdf_path)
for page in doc:
pix = page.get_pixmap()
img = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
images.append(img)
return images
except Exception as e:
print(f"Error converting PDF: {e}")
return None
def process_document(file, page_number, question, input_type):
if file is None:
return "Please upload a file."
if input_type == "PDF":
images = pdf_to_images(file)
if images is None:
return "Failed to process the PDF file."
if page_number < 1 or page_number > len(images):
return f"Invalid page number. The PDF has {len(images)} pages."
image = images[page_number - 1]
else: # Image
image = Image.open(file.name)
# prepare encoder inputs
pixel_values = processor(image, return_tensors="pt").pixel_values
# prepare decoder inputs
task_prompt = "{user_input}"
prompt = task_prompt.replace("{user_input}", question)
decoder_input_ids = processor.tokenizer(prompt, add_special_tokens=False, return_tensors="pt").input_ids
# generate answer
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=1,
bad_words_ids=[[processor.tokenizer.unk_token_id]],
return_dict_in_generate=True,
)
# postprocess
sequence = processor.batch_decode(outputs.sequences)[0]
sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
sequence = re.sub(r"<.*?>", "", sequence, count=1).strip() # remove first task start token
return processor.token2json(sequence)
def update_page_preview(file, page_number, input_type):
if file is None:
return None
if input_type == "PDF":
images = pdf_to_images(file)
if images is None or page_number < 1 or page_number > len(images):
return None
return images[page_number - 1]
else: # Image
return Image.open(file.name)
description = "Gradio Demo for Model-V3, an instance of `VisionEncoderDecoderModel` fine-tuned on DocVQA (document visual question answering). To use it, upload a PDF or image file, select a page number (for PDF), type a question, and click 'submit'."
article = "<p style='text-align: center'>Model-V3</p>"
with gr.Blocks() as demo:
gr.Markdown("# Demo: Model-V3 for Document Analysis")
gr.Markdown(description)
with gr.Row():
with gr.Column(scale=1):
input_type = gr.Radio(["PDF", "Image"], label="Input Type", value="PDF")
file_input = gr.File(label="Upload File")
page_slider = gr.Slider(minimum=1, maximum=1, value=1, step=1, label="Page Number (PDF only)")
with gr.Column(scale=2):
page_preview = gr.Image(label="Page/Image Preview")
question_input = gr.Textbox(label="Question")
submit_button = gr.Button("Submit")
output = gr.JSON(label="Output")
def update_interface(file, input_type):
if file is None:
return gr.Slider(visible=False, minimum=1, maximum=1, value=1, step=1, label="Page Number (PDF only)"), None
if input_type == "PDF":
images = pdf_to_images(file)
if images is None:
return gr.Slider(visible=False, minimum=1, maximum=1, value=1, step=1, label="Page Number (PDF only)"), None
return (
gr.Slider(visible=True, minimum=1, maximum=len(images), value=1, step=1, label="Page Number (PDF only)"),
images[0] # Show the first page by default
)
else: # Image
return gr.Slider(visible=False, minimum=1, maximum=1, value=1, step=1, label="Page Number (PDF only)"), Image.open(file.name)
input_type.change(lambda x: gr.File(label="Upload File", file_types=[".pdf"] if x == "PDF" else ["image/*"]), inputs=[input_type], outputs=[file_input])
file_input.change(update_interface, inputs=[file_input, input_type], outputs=[page_slider, page_preview])
page_slider.change(update_page_preview, inputs=[file_input, page_slider, input_type], outputs=[page_preview])
submit_button.click(process_document, inputs=[file_input, page_slider, question_input, input_type], outputs=[output])
demo.launch() |