Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import google.generativeai as genai
|
3 |
+
from PIL import Image
|
4 |
+
|
5 |
+
def extract_malayalam_text(image, api_key):
|
6 |
+
"""
|
7 |
+
Extract handwritten Malayalam text from an image using Gemini 2.0 Flash
|
8 |
+
|
9 |
+
Args:
|
10 |
+
image (PIL.Image): Uploaded image
|
11 |
+
api_key (str): Google AI Studio API Key
|
12 |
+
|
13 |
+
Returns:
|
14 |
+
str: Extracted text from the image
|
15 |
+
"""
|
16 |
+
# Validate inputs
|
17 |
+
if image is None:
|
18 |
+
return "Please upload an image first."
|
19 |
+
|
20 |
+
if not api_key:
|
21 |
+
return "Please provide a valid API key."
|
22 |
+
|
23 |
+
try:
|
24 |
+
# Configure the Gemini API directly with the provided key
|
25 |
+
genai.configure(api_key=api_key)
|
26 |
+
|
27 |
+
# Use Gemini 2.0 Flash model
|
28 |
+
model = genai.GenerativeModel('gemini-2.0-flash')
|
29 |
+
|
30 |
+
# Generate content from the image
|
31 |
+
response = model.generate_content(
|
32 |
+
[
|
33 |
+
"You are an expert in extracting handwritten Malayalam text from grocery lists. "
|
34 |
+
"Carefully transcribe each item in Malayalam script. "
|
35 |
+
"Provide a clear, accurate list of items. "
|
36 |
+
"If the text is unclear, mention potential uncertainties.",
|
37 |
+
image
|
38 |
+
],
|
39 |
+
generation_config=genai.types.GenerationConfig(
|
40 |
+
temperature=0.2, # Low temperature for precise extraction
|
41 |
+
max_output_tokens=300 # Adjust based on expected list length
|
42 |
+
)
|
43 |
+
)
|
44 |
+
|
45 |
+
# Return the extracted text
|
46 |
+
return response.text
|
47 |
+
|
48 |
+
except Exception as e:
|
49 |
+
return f"An error occurred: {str(e)}"
|
50 |
+
|
51 |
+
def create_malayalam_ocr_interface():
|
52 |
+
"""
|
53 |
+
Create Gradio interface for Malayalam OCR
|
54 |
+
"""
|
55 |
+
with gr.Blocks() as demo:
|
56 |
+
# Title and description
|
57 |
+
gr.Markdown("# Malayalam Handwritten Text Extractor")
|
58 |
+
gr.Markdown("Upload a handwritten Malayalam grocery list image for text extraction.")
|
59 |
+
|
60 |
+
# API Key input
|
61 |
+
api_key_input = gr.Textbox(
|
62 |
+
label="Google AI Studio API Key",
|
63 |
+
type="password",
|
64 |
+
placeholder="Enter your Gemini API key"
|
65 |
+
)
|
66 |
+
|
67 |
+
# Image upload component
|
68 |
+
image_input = gr.Image(
|
69 |
+
type="pil",
|
70 |
+
label="Upload Malayalam Grocery List Image"
|
71 |
+
)
|
72 |
+
|
73 |
+
# Extract button
|
74 |
+
extract_btn = gr.Button("Extract Text")
|
75 |
+
|
76 |
+
# Output text area
|
77 |
+
output_text = gr.Textbox(
|
78 |
+
label="Extracted Malayalam Text",
|
79 |
+
lines=10
|
80 |
+
)
|
81 |
+
|
82 |
+
# Connect the components
|
83 |
+
extract_btn.click(
|
84 |
+
fn=extract_malayalam_text,
|
85 |
+
inputs=[image_input, api_key_input],
|
86 |
+
outputs=output_text
|
87 |
+
)
|
88 |
+
|
89 |
+
return demo
|
90 |
+
|
91 |
+
def main():
|
92 |
+
# Create and launch the Gradio interface
|
93 |
+
demo = create_malayalam_ocr_interface()
|
94 |
+
demo.launch(
|
95 |
+
share=True, # Create a public link
|
96 |
+
debug=True # Show detailed errors
|
97 |
+
)
|
98 |
+
|
99 |
+
if __name__ == "__main__":
|
100 |
+
main()
|