SmartID Masking

AI-powered ID card detection and sensitive information masking

Demo Link:

Github Repo Url: Link

The Challenge

Sensitive information on ID cards needs to be protected to prevent identity theft and privacy breaches. Manual masking is time-consuming and prone to errors.

The Solution

We developed an AI-powered web app that detects ID card types and masks sensitive information automatically. The platform enhances privacy protection and streamlines the process of securing personal data on ID cards.

Tech Mastery Showcase

Flask

Used for building the backend API and serving the web application.

Python

Employed for data processing, machine learning model development, and backend logic implementation.

OpenCV

Utilized for image processing and manipulation, ensuring accurate detection and masking of sensitive information.

YOLO

Implemented for object detection, providing accurate identification of ID card types and sensitive fields.

JavaScript

Used for front-end interactivity and handling user inputs.

HTML & CSS

Integrated for designing the web interface and ensuring a responsive user experience.

Innovative Logic & Implementation

Data Preprocessing and Model Loading

Loaded pre-trained YOLO models for ID card type detection and sensitive information extraction. Prepared the input images for processing.

1import cv2
2    import numpy as np
3    from ultralytics import YOLO
4    
5    # Load models
6    id_card_model = YOLO('model_weights/id_card_detection.pt')
7    aadhaar_pid_model = YOLO('model_weights/aadhaar_pid_detection.pt')
8    pan_pid_model = YOLO('model_weights/pan_pid_detection.pt')
9    voter_pid_model = YOLO('model_weights/voter_pid_detection.pt')
10    passport_pid_model = YOLO('model_weights/passport_pid_detection.pt')
11    
12    # Read and preprocess image
13    image = cv2.imread('path/to/image.jpg')
14    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

ID Card Type Detection and PID Extraction

Detected the type of ID card and extracted sensitive information using the appropriate model. Masked the selected fields to protect privacy.

1def detect_card_type(image):
2    results = id_card_model(image, conf=0.25)[0]
3    detections = [
4        [*r[:4], id_card_model.names[int(r[5])]]
5        for r in results.boxes.data.tolist()
6    ]
7    return detections
8
9def extract_pids(image, model):
10    results = model(image, conf=0.25)[0]
11    detections = [
12        [*r[:4], model.names[int(r[5])]]
13        for r in results.boxes.data.tolist()
14    ]
15    return detections
16
17def process_image(image):
18    card_type_detections = detect_card_type(image)
19    # Further processing based on detected card type

Overcoming Challenges

Accurate ID Card Type Detection

Identifying various types of ID cards with high accuracy was challenging due to variations in card designs.

Solution:

Used pre-trained YOLO models and fine-tuned them with additional training data to improve detection accuracy.

Efficient Sensitive Information Masking

Masking sensitive information accurately without affecting other parts of the ID card was crucial.

Solution:

Implemented precise bounding box detection and masking techniques using OpenCV.

Creating a User-Friendly Web Interface

Ensuring the web application was intuitive and easy to use for non-technical users.

Solution:

Designed a clean and responsive UI with HTML, CSS, and JavaScript, providing clear instructions and feedback.

Key Learnings & Growth

🚀
Gained expertise in object detection models for ID card type identification.
🚀
Enhanced skills in image processing and manipulation using OpenCV.
🚀
Improved backend development with Flask for handling API requests and responses.
🚀
Developed a responsive and interactive front-end interface for user engagement.