MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202621072881 A) filed by Symbiosis International Deemed University on June 11, 2026, for An Intelligent System For Document Forgery Detection Using Ensemble Convolutional Neural Networks With Domain Adaptation.
Inventors include Dr. Harshala Shingne; Bhargav Sor; Aastha Shukla; and Arpita Mahakalkar.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT AN INTELLIGENT SYSTEM FOR DOCUMENT FORGERY DETECTION USING ENSEMBLE CONVOLUTIONAL NEURAL NETWORKS WITH DOMAIN ADAPTATION The present invention discloses an intelligent system (100) and method for automated document forgery detection using ensemble convolutional neural networks with domain adaptation analysis. The system comprises an image acquisition module (110), a preprocessing module (120) for image standardization including resizing, normalization, and label encoding, a multi-model feature extraction and classification engine (130) comprising a ResNet model (131), a MobileNetV2 model (132), and an EfficientNet model (133) each initialized with pretrained ImageNet weights and fine-tuned through transfer learning, an ensemble prediction module (140) for aggregating model predictions through weighted averaging, a performance evaluation module (150) computing accuracy, precision, recall, F1-score, and AUC metrics, and a domain adaptation analysis module (160) evaluating performance degradation caused by domain shift when transitioning from generic benchmark datasets to domain-specific document datasets. The system outputs a binary classification indicating whether the input document is genuine or forged, accompanied by confidence scores and domain shift metrics. [
Disclaimer: Curated by HT Syndication.
