Upload 3 files
Browse files- app.py +83 -0
- background_image.png +0 -0
- requirements.txt +7 -0
app.py
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
### MULTI-LANGUAGE INVOICE EXTRACTOR USING GEMINI-PRO
|
| 2 |
+
|
| 3 |
+
from dotenv import load_dotenv
|
| 4 |
+
load_dotenv() # load the environment variables
|
| 5 |
+
|
| 6 |
+
import streamlit as st
|
| 7 |
+
import os
|
| 8 |
+
import base64
|
| 9 |
+
from io import BytesIO
|
| 10 |
+
from PIL import Image
|
| 11 |
+
import google.generativeai as genai
|
| 12 |
+
|
| 13 |
+
genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
|
| 14 |
+
|
| 15 |
+
## Function to load gemini-pro-vision
|
| 16 |
+
model= genai.GenerativeModel("gemini-pro-vision")
|
| 17 |
+
|
| 18 |
+
def get_response(input, image, prompt):
|
| 19 |
+
response = model.generate_content([input, image[0], prompt]) # gemini-pro always takes input in form of a list.
|
| 20 |
+
# first input always describes or instructs how the model should behave.
|
| 21 |
+
return response
|
| 22 |
+
|
| 23 |
+
def input_image_details(uploaded_file):
|
| 24 |
+
if uploaded_file is not None:
|
| 25 |
+
# Read the file into bytes
|
| 26 |
+
bytes_data = uploaded_file.getvalue()
|
| 27 |
+
|
| 28 |
+
image_parts = [
|
| 29 |
+
{
|
| 30 |
+
"mime_type": uploaded_file.type,
|
| 31 |
+
"data": bytes_data
|
| 32 |
+
}
|
| 33 |
+
]
|
| 34 |
+
return image_parts
|
| 35 |
+
else:
|
| 36 |
+
raise FileNotFoundError("File Not Uploaded")
|
| 37 |
+
|
| 38 |
+
## initialize the streamlit app
|
| 39 |
+
|
| 40 |
+
st.set_page_config(page_title="MultiLanguage-InvoiceExtractor")
|
| 41 |
+
|
| 42 |
+
# Function to set a background image
|
| 43 |
+
def set_background(image_file):
|
| 44 |
+
with open(image_file, "rb") as image:
|
| 45 |
+
b64_image = base64.b64encode(image.read()).decode("utf-8")
|
| 46 |
+
css = f"""
|
| 47 |
+
<style>
|
| 48 |
+
.stApp {{
|
| 49 |
+
background: url(data:image/png;base64,{b64_image});
|
| 50 |
+
background-size: cover;
|
| 51 |
+
background-position: centre;
|
| 52 |
+
backgroun-repeat: no-repeat;
|
| 53 |
+
}}
|
| 54 |
+
</style>
|
| 55 |
+
"""
|
| 56 |
+
st.markdown(css, unsafe_allow_html=True)
|
| 57 |
+
|
| 58 |
+
# Set the background image
|
| 59 |
+
set_background("background_image.png")
|
| 60 |
+
|
| 61 |
+
st.header("MultiLanguage - Invoice Extractor 📄🤖")
|
| 62 |
+
|
| 63 |
+
input = st.text_input("Input Prompt: ", key="input")
|
| 64 |
+
uploaded_file = st.file_uploader("Choose an image...", type=['png', 'jpg', 'jpeg'])
|
| 65 |
+
image= ""
|
| 66 |
+
if uploaded_file is not None:
|
| 67 |
+
image = Image.open(uploaded_file)
|
| 68 |
+
st.image(image, caption="Image Uploaded.", use_column_width = True)
|
| 69 |
+
|
| 70 |
+
submit = st.button("Tell me about the invoice")
|
| 71 |
+
|
| 72 |
+
input_prompt = """
|
| 73 |
+
You are an expert in understanding invoices. We will upload an image as invoice.
|
| 74 |
+
You will have to analyze the invoice image provided to you carefully and answer
|
| 75 |
+
any questions asked related to the invoice image.
|
| 76 |
+
"""
|
| 77 |
+
|
| 78 |
+
# if submit button is clicked
|
| 79 |
+
if submit:
|
| 80 |
+
image_data = input_image_details(uploaded_file)
|
| 81 |
+
response = get_response(input_prompt, image_data, input)
|
| 82 |
+
st.subheader("Response : ")
|
| 83 |
+
st.write(response.text)
|
background_image.png
ADDED
|
requirements.txt
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
streamlit
|
| 2 |
+
google-generativeai
|
| 3 |
+
python-dotenv
|
| 4 |
+
langchain-community
|
| 5 |
+
langchain
|
| 6 |
+
PyPDF2
|
| 7 |
+
chromadb
|