Spaces:
No application file
No application file
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import AutoModel, AutoTokenizer
|
3 |
+
import os
|
4 |
+
import re
|
5 |
+
|
6 |
+
# Load the model and tokenizer from local path
|
7 |
+
# Assuming your model and tokenizer are stored in '/content/model' directory in Colab
|
8 |
+
model_path = 'pranavdaware/web_ocr'
|
9 |
+
|
10 |
+
# Load the model and tokenizer from the local directory
|
11 |
+
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
|
12 |
+
model = AutoModel.from_pretrained(model_path, trust_remote_code=True, low_cpu_mem_usage=True, device_map='cuda', use_safetensors=True)
|
13 |
+
model = model.eval().cuda()
|
14 |
+
|
15 |
+
# Function to extract text using OCR
|
16 |
+
def ocr_processing(image_file):
|
17 |
+
try:
|
18 |
+
# Perform OCR on the uploaded image
|
19 |
+
result = model.chat(tokenizer, image_file, ocr_type='ocr')
|
20 |
+
return result
|
21 |
+
except Exception as e:
|
22 |
+
return str(e)
|
23 |
+
|
24 |
+
# Function to search for keywords in extracted text
|
25 |
+
def search_keyword(ocr_text, keyword):
|
26 |
+
try:
|
27 |
+
# Use regex to search for the keyword and highlight matches
|
28 |
+
matches = re.findall(rf"({keyword})", ocr_text, re.IGNORECASE)
|
29 |
+
if matches:
|
30 |
+
highlighted_text = re.sub(rf"({keyword})", r'<mark>\1</mark>', ocr_text, flags=re.IGNORECASE)
|
31 |
+
return highlighted_text
|
32 |
+
else:
|
33 |
+
return f"No matches found for '{keyword}' in the extracted text."
|
34 |
+
except Exception as e:
|
35 |
+
return str(e)
|
36 |
+
|
37 |
+
# Gradio interface
|
38 |
+
def main():
|
39 |
+
# Gradio app layout
|
40 |
+
with gr.Blocks() as demo:
|
41 |
+
gr.Markdown("# OCR and Keyword Search Application")
|
42 |
+
|
43 |
+
with gr.Row():
|
44 |
+
with gr.Column():
|
45 |
+
image_input = gr.Image(type="filepath", label="Upload your image")
|
46 |
+
keyword_input = gr.Textbox(label="Enter keyword to search")
|
47 |
+
ocr_output = gr.Textbox(label="OCR Output")
|
48 |
+
search_output = gr.HTML(label="Search Results")
|
49 |
+
|
50 |
+
# Button for OCR processing
|
51 |
+
process_button = gr.Button("Process Image for OCR")
|
52 |
+
|
53 |
+
# Connect the OCR processing function to the button
|
54 |
+
process_button.click(
|
55 |
+
fn=ocr_processing,
|
56 |
+
inputs=image_input,
|
57 |
+
outputs=ocr_output
|
58 |
+
)
|
59 |
+
|
60 |
+
# Button for keyword search
|
61 |
+
search_button = gr.Button("Search Keyword in OCR Text")
|
62 |
+
|
63 |
+
# Connect the search function to the button
|
64 |
+
search_button.click(
|
65 |
+
fn=search_keyword,
|
66 |
+
inputs=[ocr_output, keyword_input],
|
67 |
+
outputs=search_output
|
68 |
+
)
|
69 |
+
|
70 |
+
# Launch the Gradio demo
|
71 |
+
demo.launch()
|
72 |
+
|
73 |
+
if __name__ == "__main__":
|
74 |
+
main()
|