Spaces:
Runtime error
Runtime error
Commit
•
a4d7a24
1
Parent(s):
06d7744
Update app.py
Browse files
app.py
CHANGED
@@ -1,5 +1,3 @@
|
|
1 |
-
import os
|
2 |
-
|
3 |
import os
|
4 |
os.system('pip install git+https://github.com/huggingface/transformers.git --upgrade')
|
5 |
os.system('pip install pyyaml==5.1')
|
@@ -8,29 +6,23 @@ os.system('pip install torch==1.8.0+cu101 torchvision==0.9.0+cu101 -f https://do
|
|
8 |
|
9 |
# install detectron2 that matches pytorch 1.8
|
10 |
# See https://detectron2.readthedocs.io/tutorials/install.html for instructions
|
11 |
-
|
12 |
|
13 |
## install PyTesseract
|
14 |
os.system('pip install -q pytesseract')
|
15 |
|
16 |
-
import gradio as gr
|
17 |
-
import torch
|
18 |
-
import PIL
|
19 |
import gradio as gr
|
20 |
import numpy as np
|
21 |
-
from transformers import
|
22 |
from datasets import load_dataset
|
23 |
from PIL import Image, ImageDraw, ImageFont
|
24 |
|
25 |
-
|
26 |
-
|
27 |
-
processor = LayoutLMv3Processor.from_pretrained("Ammar-alhaj-ali/LayoutLMv3-Fine-Tuning-FUNSD")
|
28 |
-
model = LayoutLMv3ForTokenClassification.from_pretrained("Ammar-alhaj-ali/LayoutLMv3-Fine-Tuning-FUNSD")
|
29 |
|
30 |
# load image example
|
31 |
dataset = load_dataset("nielsr/funsd", split="test")
|
32 |
image = Image.open(dataset[0]["image_path"]).convert("RGB")
|
33 |
-
#image = Image.open("./invoice.png")
|
34 |
image.save("document.png")
|
35 |
# define id2label, label2color
|
36 |
labels = dataset.features['ner_tags'].feature.names
|
@@ -80,17 +72,12 @@ def process_image(image):
|
|
80 |
|
81 |
return image
|
82 |
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2012.14740' target='_blank'>LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding</a> | <a href='https://github.com/microsoft/unilm' target='_blank'>Github Repo</a></p>"
|
87 |
examples =[['img1.png']]
|
88 |
|
89 |
-
css = ".
|
90 |
-
#css = "@media screen and (max-width: 600px) { .output_image, .input_image {height:20rem !important; width: 100% !important;} }"
|
91 |
-
# css = ".output_image, .input_image {height: 600px !important}"
|
92 |
-
|
93 |
-
css = ".image-preview {height: auto !important;}"
|
94 |
|
95 |
iface = gr.Interface(fn=process_image,
|
96 |
inputs=gr.inputs.Image(type="pil"),
|
@@ -99,6 +86,5 @@ iface = gr.Interface(fn=process_image,
|
|
99 |
description=description,
|
100 |
article=article,
|
101 |
examples=examples,
|
102 |
-
css=css
|
103 |
-
enable_queue=True)
|
104 |
iface.launch(debug=True)
|
|
|
|
|
|
|
1 |
import os
|
2 |
os.system('pip install git+https://github.com/huggingface/transformers.git --upgrade')
|
3 |
os.system('pip install pyyaml==5.1')
|
|
|
6 |
|
7 |
# install detectron2 that matches pytorch 1.8
|
8 |
# See https://detectron2.readthedocs.io/tutorials/install.html for instructions
|
9 |
+
os.system('pip install -q detectron2 -f https://dl.fbaipublicfiles.com/detectron2/wheels/cu101/torch1.8/index.html')
|
10 |
|
11 |
## install PyTesseract
|
12 |
os.system('pip install -q pytesseract')
|
13 |
|
|
|
|
|
|
|
14 |
import gradio as gr
|
15 |
import numpy as np
|
16 |
+
from transformers import LayoutLMv2Processor, LayoutLMv2ForTokenClassification
|
17 |
from datasets import load_dataset
|
18 |
from PIL import Image, ImageDraw, ImageFont
|
19 |
|
20 |
+
processor = LayoutLMv2Processor.from_pretrained("microsoft/layoutlmv2-base-uncased")
|
21 |
+
model = LayoutLMv2ForTokenClassification.from_pretrained("nielsr/layoutlmv2-finetuned-funsd")
|
|
|
|
|
22 |
|
23 |
# load image example
|
24 |
dataset = load_dataset("nielsr/funsd", split="test")
|
25 |
image = Image.open(dataset[0]["image_path"]).convert("RGB")
|
|
|
26 |
image.save("document.png")
|
27 |
# define id2label, label2color
|
28 |
labels = dataset.features['ner_tags'].feature.names
|
|
|
72 |
|
73 |
return image
|
74 |
|
75 |
+
title = "Interactive demo: LayoutLMv2"
|
76 |
+
description = "Demo for Microsoft's LayoutLMv2, a Transformer for state-of-the-art document image understanding tasks. This particular model is fine-tuned on FUNSD, a dataset of manually annotated forms. It annotates the words into QUESTION/ANSWER/HEADER/OTHER. To use it, simply upload an image or use the example image below. Results will show up in a few seconds."
|
77 |
+
article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2012.14740'>LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding</a> | <a href='https://github.com/microsoft/unilm'>Github Repo</a></p>"
|
|
|
78 |
examples =[['img1.png']]
|
79 |
|
80 |
+
css = """.output_image, .input_image {height: 600px !important}"""
|
|
|
|
|
|
|
|
|
81 |
|
82 |
iface = gr.Interface(fn=process_image,
|
83 |
inputs=gr.inputs.Image(type="pil"),
|
|
|
86 |
description=description,
|
87 |
article=article,
|
88 |
examples=examples,
|
89 |
+
css=css)
|
|
|
90 |
iface.launch(debug=True)
|