Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -80,10 +80,14 @@ def polish_text_with_ai(cleaned_text):
|
|
| 80 |
return polished_text
|
| 81 |
|
| 82 |
# Extract text using GOT
|
|
|
|
|
|
|
| 83 |
def extract_text_got(image_file, model, tokenizer):
|
| 84 |
return model.chat(tokenizer, image_file, ocr_type='ocr')
|
| 85 |
|
| 86 |
# Extract text using Qwen
|
|
|
|
|
|
|
| 87 |
def extract_text_qwen(image_file, model, processor):
|
| 88 |
try:
|
| 89 |
image = Image.open(image_file).convert('RGB')
|
|
@@ -103,14 +107,14 @@ def extract_text_qwen(image_file, model, processor):
|
|
| 103 |
# Function to highlight the keyword in the text
|
| 104 |
|
| 105 |
|
| 106 |
-
def highlight_text(cleaned_text,start,end):
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
)
|
| 114 |
|
| 115 |
|
| 116 |
# Title and UI
|
|
@@ -129,7 +133,8 @@ uploaded_file = st.sidebar.file_uploader(
|
|
| 129 |
|
| 130 |
# Input from clipboard
|
| 131 |
# Paste image button
|
| 132 |
-
image_data =
|
|
|
|
| 133 |
if image_data is not None:
|
| 134 |
clipboard_use = True
|
| 135 |
header, encoded = image_data.split(",", 1)
|
|
@@ -217,6 +222,6 @@ if uploaded_file:
|
|
| 217 |
len = search_query.length
|
| 218 |
end = index + len
|
| 219 |
if index != -1:
|
| 220 |
-
highlight_text(cleaned_text,start,end)
|
| 221 |
else:
|
| 222 |
st.write("No Search Found.")
|
|
|
|
| 80 |
return polished_text
|
| 81 |
|
| 82 |
# Extract text using GOT
|
| 83 |
+
|
| 84 |
+
|
| 85 |
def extract_text_got(image_file, model, tokenizer):
|
| 86 |
return model.chat(tokenizer, image_file, ocr_type='ocr')
|
| 87 |
|
| 88 |
# Extract text using Qwen
|
| 89 |
+
|
| 90 |
+
|
| 91 |
def extract_text_qwen(image_file, model, processor):
|
| 92 |
try:
|
| 93 |
image = Image.open(image_file).convert('RGB')
|
|
|
|
| 107 |
# Function to highlight the keyword in the text
|
| 108 |
|
| 109 |
|
| 110 |
+
def highlight_text(cleaned_text, start, end):
|
| 111 |
+
text_highlighter(
|
| 112 |
+
text=cleaned_text,
|
| 113 |
+
labels=[("KEYWORD", "#0000FF")],
|
| 114 |
+
annotations=[
|
| 115 |
+
{"start": start, "end": end, "tag": "KEYWORD"},
|
| 116 |
+
],
|
| 117 |
+
)
|
| 118 |
|
| 119 |
|
| 120 |
# Title and UI
|
|
|
|
| 133 |
|
| 134 |
# Input from clipboard
|
| 135 |
# Paste image button
|
| 136 |
+
image_data = paste(
|
| 137 |
+
label="Paste From Clipboard", key="image_clipboard")
|
| 138 |
if image_data is not None:
|
| 139 |
clipboard_use = True
|
| 140 |
header, encoded = image_data.split(",", 1)
|
|
|
|
| 222 |
len = search_query.length
|
| 223 |
end = index + len
|
| 224 |
if index != -1:
|
| 225 |
+
highlight_text(cleaned_text, start, end)
|
| 226 |
else:
|
| 227 |
st.write("No Search Found.")
|