usmanabbasi/Ekyc_Using_Computer_Vision

1import streamlit as st2from PIL import Image3import numpy as np4import easyocr5import pandas as pd6import base647import re8from datetime import datetime, timedelta9 10def process_image(image):11    reader = easyocr.Reader(['en'], gpu=False)12    img_np = np.array(image)13    result = reader.readtext(img_np)14 15    extracted_data = {16        "Name": None,17        "Father Name": None,18        "Gender": None,19        "Country of Stay": "Pakistan",20        "Identity Number": None,21        "Date of Birth": None,22        "Date of Issue": None,23        "Date of Expiry": None24    }25 26    for i, detection in enumerate(result):27        text = detection[1].strip()28        if "name" in text.lower() and not "father" in text.lower():29            extracted_data["Name"] = result[i+1][1].strip() if i+1 < len(result) else None30        elif "father" in text.lower():31            extracted_data["Father Name"] = result[i+1][1].strip() if i+1 < len(result) else None32        elif text.lower() in ["m", "f"]:33            extracted_data["Gender"] = text.upper()34        elif re.match(r'\d{5}-\d{7}-\d', text):35            extracted_data["Identity Number"] = text36        elif re.match(r'\d{2}\.\d{2}\.\d{4}', text):37            if extracted_data["Date of Birth"] is None:38                extracted_data["Date of Birth"] = text39            elif extracted_data["Date of Issue"] is None:40                extracted_data["Date of Issue"] = text41 42    if extracted_data["Date of Issue"] and not extracted_data["Date of Expiry"]:43        try:44            date_of_issue = datetime.strptime(extracted_data["Date of Issue"], "%d.%m.%Y")45            date_of_expiry = date_of_issue.replace(year=date_of_issue.year + 10)46            extracted_data["Date of Expiry"] = date_of_expiry.strftime("%d.%m.%Y")47        except ValueError:48            pass49 50    return extracted_data51 52def display_table(extracted_data):53    fields = ["Name", "Father Name", "Gender", "Country of Stay", "Identity Number", "Date of Birth", "Date of Issue", "Date of Expiry"]54    values = [extracted_data[field] if extracted_data[field] else "" for field in fields]55    df = pd.DataFrame(list(zip(fields, values)), columns=['Field', 'Value'])56    st.dataframe(df)57 58def get_csv_download_link(df):59    csv = df.to_csv(index=False)60    b64 = base64.b64encode(csv.encode()).decode()61    href = f'<a href="data:file/csv;base64,{b64}" download="extracted_data.csv">Download CSV File</a>'62    return href63 64def data_extraction_page():65    st.title('ID Card Text Extraction')66 67    uploaded_file = st.file_uploader("Upload an image of your ID card to Extract Data", type=["jpg", "jpeg", "png"])68 69    if uploaded_file is not None:70        image = Image.open(uploaded_file)71        st.image(image, caption='Wait...! We Are Extracting Data For You', use_column_width=True)72        extracted_data = process_image(image)73        display_table(extracted_data)74 75        st.markdown(get_csv_download_link(pd.DataFrame(list(extracted_data.items()), columns=['Field', 'Value'])), unsafe_allow_html=True)76