MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202621075919 A) filed by Symbiosis International Deemed University on June 18, 2026, for Multi-Modal Ai-Based System And Method For Automated Detection Of Forged Documents Using Explainable Forensic Analysis.
Inventors include Dr. Deepak Suresh Asudani; Mukund Kuthe; Rushi Parhad; and Sharvayu Zade.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: ABSTRACT MULTI-MODAL AI-BASED SYSTEM AND METHOD FOR AUTOMATED DETECTION OF FORGED DOCUMENTS USING EXPLAINABLE FORENSIC ANALYSIS The present invention discloses a multi-modal AI-based system (100) and method for automated detection of forged documents. The system (100) comprises a document input and preprocessing unit (110), a CNN-based visual forgery detection module (120) employing MobileNetV2 with Error Level Analysis enhancement, an OCR-based semantic validation module (130), a signature verification module (140), a GNN-based structural analysis module (150) for document layout reasoning, a digital verification module (160) with blockchain-backed hash verification, a weighted fusion decision engine (170) for aggregating multi- module confidence scores, and an explainability engine (180) using Grad-CAM visualization. The system (100) integrates visual, textual, structural, and digital modalities through the weighted fusion engine (170), which assesses cross-module agreement and applies adaptive thresholds optimized for fraud prevention. The system (100) outputs a final authenticity verdict with confidence score, tampering reasons, and interpretable visual evidence, providing a legally defensible solution for document verification in financial, governmental, and legal applications. [
Disclaimer: Curated by HT Syndication.
