MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641079717 A) filed by Sona College Of Technology on June 29, 2026, for Camera-Assisted Neuro-Symbolic System For Automated Form Understanding And Data Population.
Inventors include Selvanathan N; Murali M; Karthick K; Dhanushraj T; Dharanesh V; and Inbasarathi J.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: ABSTRACT A camera-assisted neuro-symbolic system for automated form understanding and data population is disclosed. The system comprises an optical hardware module including one or more cameras configured to capture images of physical forms and documents. A preprocessing and perception module performs image enhancement, alignment, region detection, and confidence-based text extraction to generate uncertainty-aware evidence. A Document Structure Graph (DSG) builder constructs a graph representation of document elements and their spatial, semantic, and logical relationships. A neural inference engine analyzes the DSG to classify fields and infer missing or uncertain information. A symbolic reasoning module validates extracted and inferred data using predefined constraints and dependency rules, while a probabilistic inference layer resolves ambiguities through confidence-weighted reasoning. A self-evolving learning module updates inference and validation parameters based on feedback and correction data. A form population and validation module automatically fills digital forms with validated information. The invention enables accurate, explainable, adaptive, and structure-aware document automation across diverse document layouts.
Disclaimer: Curated by HT Syndication.
