Spaces:
Sleeping
Sleeping
Create app_v3.txt
Browse files- app_v3.txt +320 -0
app_v3.txt
ADDED
|
@@ -0,0 +1,320 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import pandas as pd
|
| 3 |
+
import numpy as np
|
| 4 |
+
import easyocr
|
| 5 |
+
import streamlit as st
|
| 6 |
+
from annotated_text import annotated_text
|
| 7 |
+
from streamlit_option_menu import option_menu
|
| 8 |
+
from sentiment_analysis import SentimentAnalysis
|
| 9 |
+
from keyword_extraction import KeywordExtractor
|
| 10 |
+
from part_of_speech_tagging import POSTagging
|
| 11 |
+
from emotion_detection import EmotionDetection
|
| 12 |
+
from named_entity_recognition import NamedEntityRecognition
|
| 13 |
+
from Object_Detector import ObjectDetector
|
| 14 |
+
from OCR_Detector import OCRDetector
|
| 15 |
+
import PIL
|
| 16 |
+
from PIL import Image
|
| 17 |
+
from PIL import ImageColor
|
| 18 |
+
from PIL import ImageDraw
|
| 19 |
+
from PIL import ImageFont
|
| 20 |
+
import time
|
| 21 |
+
|
| 22 |
+
# Imports de Object Detection
|
| 23 |
+
import tensorflow as tf
|
| 24 |
+
import tensorflow_hub as hub
|
| 25 |
+
# Load compressed models from tensorflow_hub
|
| 26 |
+
os.environ['TFHUB_MODEL_LOAD_FORMAT'] = 'COMPRESSED'
|
| 27 |
+
import matplotlib.pyplot as plt
|
| 28 |
+
import matplotlib as mpl
|
| 29 |
+
# For drawing onto the image.
|
| 30 |
+
import numpy as np
|
| 31 |
+
from tensorflow.python.ops.numpy_ops import np_config
|
| 32 |
+
np_config.enable_numpy_behavior()
|
| 33 |
+
|
| 34 |
+
import torch
|
| 35 |
+
import librosa
|
| 36 |
+
from models import infere_speech_emotion, infere_text_emotion, infere_voice2text
|
| 37 |
+
|
| 38 |
+
st.set_page_config(layout="wide")
|
| 39 |
+
|
| 40 |
+
hide_streamlit_style = """
|
| 41 |
+
<style>
|
| 42 |
+
#MainMenu {visibility: hidden;}
|
| 43 |
+
footer {visibility: hidden;}
|
| 44 |
+
</style>
|
| 45 |
+
"""
|
| 46 |
+
st.markdown(hide_streamlit_style, unsafe_allow_html=True)
|
| 47 |
+
|
| 48 |
+
@st.cache_resource
|
| 49 |
+
def load_sentiment_model():
|
| 50 |
+
return SentimentAnalysis()
|
| 51 |
+
|
| 52 |
+
@st.cache_resource
|
| 53 |
+
def load_keyword_model():
|
| 54 |
+
return KeywordExtractor()
|
| 55 |
+
|
| 56 |
+
@st.cache_resource
|
| 57 |
+
def load_pos_model():
|
| 58 |
+
return POSTagging()
|
| 59 |
+
|
| 60 |
+
@st.cache_resource
|
| 61 |
+
def load_emotion_model():
|
| 62 |
+
return EmotionDetection()
|
| 63 |
+
|
| 64 |
+
@st.cache_resource
|
| 65 |
+
def load_ner_model():
|
| 66 |
+
return NamedEntityRecognition()
|
| 67 |
+
|
| 68 |
+
@st.cache_resource
|
| 69 |
+
def load_objectdetector_model():
|
| 70 |
+
return ObjectDetector()
|
| 71 |
+
|
| 72 |
+
@st.cache_resource
|
| 73 |
+
def load_ocrdetector_model():
|
| 74 |
+
return OCRDetector()
|
| 75 |
+
|
| 76 |
+
sentiment_analyzer = load_sentiment_model()
|
| 77 |
+
keyword_extractor = load_keyword_model()
|
| 78 |
+
pos_tagger = load_pos_model()
|
| 79 |
+
emotion_detector = load_emotion_model()
|
| 80 |
+
ner = load_ner_model()
|
| 81 |
+
objectdetector1 = load_objectdetector_model()
|
| 82 |
+
ocrdetector1 = load_ocrdetector_model()
|
| 83 |
+
|
| 84 |
+
def rectangle(image, result):
|
| 85 |
+
draw = ImageDraw.Draw(image)
|
| 86 |
+
for res in result:
|
| 87 |
+
top_left = tuple(res[0][0]) # top left coordinates as tuple
|
| 88 |
+
bottom_right = tuple(res[0][2]) # bottom right coordinates as tuple
|
| 89 |
+
draw.rectangle((top_left, bottom_right), outline="blue", width=2)
|
| 90 |
+
st.image(image)
|
| 91 |
+
|
| 92 |
+
example_text = "My name is Daniel: The attention to detail, swift resolution, and accuracy demonstrated by ITACA Insurance Company in Spain in handling my claim were truly impressive. This undoubtedly reflects their commitment to being a customer-centric insurance provider."
|
| 93 |
+
|
| 94 |
+
with st.sidebar:
|
| 95 |
+
image = Image.open('./itaca_logo.png')
|
| 96 |
+
st.image(image,width=150) #use_column_width=True)
|
| 97 |
+
page = option_menu(menu_title='Menu',
|
| 98 |
+
menu_icon="robot",
|
| 99 |
+
options=["Sentiment Analysis",
|
| 100 |
+
"Keyword Extraction",
|
| 101 |
+
"Part of Speech Tagging",
|
| 102 |
+
"Emotion Detection",
|
| 103 |
+
"Named Entity Recognition",
|
| 104 |
+
"Speech & Text Emotion",
|
| 105 |
+
"Object Detector",
|
| 106 |
+
"OCR Detector"],
|
| 107 |
+
icons=["chat-dots",
|
| 108 |
+
"key",
|
| 109 |
+
"tag",
|
| 110 |
+
"emoji-heart-eyes",
|
| 111 |
+
"building",
|
| 112 |
+
"book",
|
| 113 |
+
"camera",
|
| 114 |
+
"list-task"],
|
| 115 |
+
default_index=0
|
| 116 |
+
)
|
| 117 |
+
|
| 118 |
+
st.title('ITACA Insurance Core AI Module')
|
| 119 |
+
|
| 120 |
+
# Replace '20px' with your desired font size
|
| 121 |
+
font_size = '20px'
|
| 122 |
+
|
| 123 |
+
if page == "Sentiment Analysis":
|
| 124 |
+
st.header('Sentiment Analysis')
|
| 125 |
+
# st.markdown("")
|
| 126 |
+
st.write(
|
| 127 |
+
"""
|
| 128 |
+
"""
|
| 129 |
+
)
|
| 130 |
+
|
| 131 |
+
text = st.text_area("Paste text here", value=example_text)
|
| 132 |
+
|
| 133 |
+
if st.button('🔥 Run!'):
|
| 134 |
+
with st.spinner("Loading..."):
|
| 135 |
+
preds, html = sentiment_analyzer.run(text)
|
| 136 |
+
st.success('All done!')
|
| 137 |
+
st.write("")
|
| 138 |
+
st.subheader("Sentiment Predictions")
|
| 139 |
+
st.bar_chart(data=preds, width=0, height=0, use_container_width=True)
|
| 140 |
+
st.write("")
|
| 141 |
+
st.subheader("Sentiment Justification")
|
| 142 |
+
raw_html = html._repr_html_()
|
| 143 |
+
st.components.v1.html(raw_html, height=500)
|
| 144 |
+
|
| 145 |
+
elif page == "Keyword Extraction":
|
| 146 |
+
st.header('Keyword Extraction')
|
| 147 |
+
# st.markdown("")
|
| 148 |
+
st.write(
|
| 149 |
+
"""
|
| 150 |
+
"""
|
| 151 |
+
)
|
| 152 |
+
|
| 153 |
+
text = st.text_area("Paste text here", value=example_text)
|
| 154 |
+
|
| 155 |
+
max_keywords = st.slider('# of Keywords Max Limit', min_value=1, max_value=10, value=5, step=1)
|
| 156 |
+
|
| 157 |
+
if st.button('🔥 Run!'):
|
| 158 |
+
with st.spinner("Loading..."):
|
| 159 |
+
annotation, keywords = keyword_extractor.generate(text, max_keywords)
|
| 160 |
+
st.success('All done!')
|
| 161 |
+
|
| 162 |
+
if annotation:
|
| 163 |
+
st.subheader("Keyword Annotation")
|
| 164 |
+
st.write("")
|
| 165 |
+
annotated_text(*annotation)
|
| 166 |
+
st.text("")
|
| 167 |
+
|
| 168 |
+
st.subheader("Extracted Keywords")
|
| 169 |
+
st.write("")
|
| 170 |
+
df = pd.DataFrame(keywords, columns=['Extracted Keywords'])
|
| 171 |
+
csv = df.to_csv(index=False).encode('utf-8')
|
| 172 |
+
st.download_button('Download Keywords to CSV', csv, file_name='news_intelligence_keywords.csv')
|
| 173 |
+
|
| 174 |
+
data_table = st.table(df)
|
| 175 |
+
|
| 176 |
+
elif page == "Part of Speech Tagging":
|
| 177 |
+
st.header('Part of Speech Tagging')
|
| 178 |
+
# st.markdown("")
|
| 179 |
+
st.write(
|
| 180 |
+
"""
|
| 181 |
+
"""
|
| 182 |
+
)
|
| 183 |
+
|
| 184 |
+
text = st.text_area("Paste text here", value=example_text)
|
| 185 |
+
|
| 186 |
+
if st.button('🔥 Run!'):
|
| 187 |
+
with st.spinner("Loading..."):
|
| 188 |
+
preds = pos_tagger.classify(text)
|
| 189 |
+
st.success('All done!')
|
| 190 |
+
st.write("")
|
| 191 |
+
st.subheader("Part of Speech tags")
|
| 192 |
+
annotated_text(*preds)
|
| 193 |
+
st.write("")
|
| 194 |
+
st.components.v1.iframe('https://www.ling.upenn.edu/courses/Fall_2003/ling001/penn_treebank_pos.html', height=1000)
|
| 195 |
+
|
| 196 |
+
elif page == "Emotion Detection":
|
| 197 |
+
st.header('Emotion Detection')
|
| 198 |
+
# st.markdown("")
|
| 199 |
+
st.write(
|
| 200 |
+
"""
|
| 201 |
+
"""
|
| 202 |
+
)
|
| 203 |
+
|
| 204 |
+
text = st.text_area("Paste text here", value=example_text)
|
| 205 |
+
|
| 206 |
+
if st.button('🔥 Run!'):
|
| 207 |
+
with st.spinner("Loading..."):
|
| 208 |
+
preds, html = emotion_detector.run(text)
|
| 209 |
+
st.success('All done!')
|
| 210 |
+
st.write("")
|
| 211 |
+
st.subheader("Emotion Predictions")
|
| 212 |
+
st.bar_chart(data=preds, width=0, height=0, use_container_width=True)
|
| 213 |
+
raw_html = html._repr_html_()
|
| 214 |
+
st.write("")
|
| 215 |
+
st.subheader("Emotion Justification")
|
| 216 |
+
st.components.v1.html(raw_html, height=500)
|
| 217 |
+
|
| 218 |
+
elif page == "Named Entity Recognition":
|
| 219 |
+
st.header('Named Entity Recognition')
|
| 220 |
+
# st.markdown("")
|
| 221 |
+
st.write(
|
| 222 |
+
"""
|
| 223 |
+
"""
|
| 224 |
+
)
|
| 225 |
+
|
| 226 |
+
text = st.text_area("Paste text here", value=example_text)
|
| 227 |
+
|
| 228 |
+
if st.button('🔥 Run!'):
|
| 229 |
+
with st.spinner("Loading..."):
|
| 230 |
+
preds, ner_annotation = ner.classify(text)
|
| 231 |
+
st.success('All done!')
|
| 232 |
+
st.write("")
|
| 233 |
+
st.subheader("NER Predictions")
|
| 234 |
+
annotated_text(*ner_annotation)
|
| 235 |
+
st.write("")
|
| 236 |
+
st.subheader("NER Prediction Metadata")
|
| 237 |
+
st.write(preds)
|
| 238 |
+
|
| 239 |
+
elif page == "Object Detector":
|
| 240 |
+
st.header('Object Detector')
|
| 241 |
+
st.write(
|
| 242 |
+
"""
|
| 243 |
+
"""
|
| 244 |
+
)
|
| 245 |
+
|
| 246 |
+
img_file_buffer = st.file_uploader("Load an image", type=["png", "jpg", "jpeg"])
|
| 247 |
+
if img_file_buffer is not None:
|
| 248 |
+
image = np.array(Image.open(img_file_buffer))
|
| 249 |
+
|
| 250 |
+
if st.button('🔥 Run!'):
|
| 251 |
+
with st.spinner("Loading..."):
|
| 252 |
+
img, primero = objectdetector1.run_detector(image)
|
| 253 |
+
st.success('The first image detected is: ' + primero)
|
| 254 |
+
st.image(img, caption="Imagen", use_column_width=True)
|
| 255 |
+
|
| 256 |
+
elif page == "OCR Detector":
|
| 257 |
+
st.header('OCR Detector')
|
| 258 |
+
st.write(
|
| 259 |
+
"""
|
| 260 |
+
"""
|
| 261 |
+
)
|
| 262 |
+
|
| 263 |
+
file = st.file_uploader("Load an image", type=["png", "jpg", "jpeg"])
|
| 264 |
+
|
| 265 |
+
#read the csv file and display the dataframe
|
| 266 |
+
if file is not None:
|
| 267 |
+
image = Image.open(file) # read image with PIL library
|
| 268 |
+
|
| 269 |
+
if st.button('🔥 Run!'):
|
| 270 |
+
with st.spinner("Loading..."):
|
| 271 |
+
result = ocrdetector1.reader.readtext(np.array(image)) # turn image to numpy array
|
| 272 |
+
|
| 273 |
+
# collect the results in dictionary:
|
| 274 |
+
textdic_easyocr = {}
|
| 275 |
+
for idx in range(len(result)):
|
| 276 |
+
pred_coor = result[idx][0]
|
| 277 |
+
pred_text = result[idx][1]
|
| 278 |
+
pred_confidence = result[idx][2]
|
| 279 |
+
textdic_easyocr[pred_text] = {}
|
| 280 |
+
textdic_easyocr[pred_text]['pred_confidence'] = pred_confidence
|
| 281 |
+
|
| 282 |
+
# get boxes on the image
|
| 283 |
+
rectangle(image, result)
|
| 284 |
+
|
| 285 |
+
# create a dataframe which shows the predicted text and prediction confidence
|
| 286 |
+
df = pd.DataFrame.from_dict(textdic_easyocr).T
|
| 287 |
+
st.table(df)
|
| 288 |
+
elif page == "Speech & Text Emotion":
|
| 289 |
+
st.header('Speech & Text Emotion')
|
| 290 |
+
st.write(
|
| 291 |
+
"""
|
| 292 |
+
"""
|
| 293 |
+
)
|
| 294 |
+
uploaded_file = st.file_uploader("Choose an audio file", type=["mp3", "wav", "ogg"])
|
| 295 |
+
|
| 296 |
+
if uploaded_file is not None:
|
| 297 |
+
st.audio(uploaded_file, format='audio/' + uploaded_file.type.split('/')[1])
|
| 298 |
+
st.write("Audio file uploaded and playing.")
|
| 299 |
+
|
| 300 |
+
else:
|
| 301 |
+
st.write("Please upload an audio file.")
|
| 302 |
+
|
| 303 |
+
if st.button("Analysis"):
|
| 304 |
+
with st.spinner("Loading..."):
|
| 305 |
+
st.header('Results of the Audio & Text analysis:')
|
| 306 |
+
samples, sample_rate = librosa.load(uploaded_file, sr=16000)
|
| 307 |
+
p_voice2text = infere_voice2text (samples)
|
| 308 |
+
p_speechemotion = infere_speech_emotion(samples)
|
| 309 |
+
p_textemotion = infere_text_emotion(p_voice2text)
|
| 310 |
+
st.subheader("Text from the Audio:")
|
| 311 |
+
st.write(p_voice2text)
|
| 312 |
+
st.write("---")
|
| 313 |
+
st.subheader("Speech emotion:")
|
| 314 |
+
st.write(p_speechemotion)
|
| 315 |
+
st.write("---")
|
| 316 |
+
st.subheader("Text emotion:")
|
| 317 |
+
st.write(p_textemotion)
|
| 318 |
+
st.write("---")
|
| 319 |
+
|
| 320 |
+
|