MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202641046135 A) filed by Priyanak K; Pavithra Kumar Nadar; Prakash C; Praveshika P; and Sanchana N M on April 10, 2026, for Ai Powered Form Autofill.
Inventors include Priyanak K; Pavithra Kumar; Prakash C; Praveshika P; and Sanchana N M.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: In an era of digitized governance and enterprise operations, citizens and professionals are repeatedly burdened with manually filling identical information across multiple forms — a process that is error- prone, time-consuming, and frustrating. This paper presents a production-grade Al-Powered Form Autofill Web Application that leverages a multi-layer, fault-tolerant document extraction pipeline to intelligently extract structured data from identity documents such as PAN cards, Aadhaar cards, Passports, and Resumes, and automatically populate web forms, PDF forms, and Google Forms with high accuracy. The system employs a seven-layer extraction architecture incorporating image preprocessing, ensemble OCR (Tesseract, EasyOCR, PaddleOCR), document-type classification, deterministic regexbased extraction, context aware field disambiguation, cross-validation, and confidence scoring per field. An intelligent field mapping engine using label matching, fuzzy matching (RapidFuzz), embedding similarity (Sentence-BERT), and LLM fallback reasoning ensures that semantically equivalent but syntactically diverse form fields are correctly mapped. Built with a FastAPI backend and a React frontend, the system is deployable as a real SaaS product. Experimental results demonstrate nearperfect PAN extraction (99.1%), Aadhaar number extraction (98.8%), 100% Passport number extraction, and overall field mapping accuracy of 97.9% across 425 test mappings.
Disclaimer: Curated by HT Syndication.
