Targeted Demographics Extraction from Complex Identity Cards
Aadhaar Card Data Extractor is an intelligent, full-stack hybrid AI application that automates the extraction of key demographic fields (Name, Date of Birth, Gender, and Aadhaar Number) from Aadhaar cards. It utilizes a custom-trained YOLOv8 object detection model to locate text regions, applies advanced OpenCV preprocessing filters for text clarity, and runs Tesseract OCR to parse the content with high-accuracy fallbacks.
Designed with a modular FastAPI backend and a modern drag-and-drop web UI built using Vanilla JS, HTML5, and CSS3, the application processes images in a pipeline: Preprocessing -> YOLOv8 Inference -> OCR -> Structured JSON Data. It eliminates background noise by targeting specific Regions of Interest (ROI) and cropping high-resolution segments from original images.
To handle bad lighting or rotation, the backend incorporates robust fallback logic, including a full-card OCR text scan and targeted bottom-half sweeps. This hybrid approach makes it production-ready for onboarding, KYC verification, and automated data entry workflows.
Architectural Components
Core Technologies
Python
Python
FastAPI
FastAPI
YOLOv8
YOLOv8
Tesseract OCR
Tesseract OCR
HTML
HTML
CSS
CSS
JavaScript
JavaScript
Interface Design & Showcase
Project Gallery
Sleek dark mode dashboard with drag-and-drop file upload zone and side-by-side extracted data.
