MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641107433 A) filed by Kkr & Ksr Institute Of Technology And Sciences on September 07, 2026, for Unmasking The Deep Fakes: Exposing The Truths Behind Synthetic Media Using Deep Learning.
Inventors include Mr. Prof. Repudi Ramesh; Mrs. Dr. Chittineni Aruna; Mr. Gogula Mahesh Reddy; Donthu Naga Reethika; Chitla Bhavya Sri; Galla Poojitha; and Gayatri Rachumallu.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: ABSTRACT [0011] An artificial-intelligence-based deepfake detection system is disclosed that identifies manipulated facial images and videos by examining spatial and temporal artifacts introduced during synthesis. A convolutional neural network extracts high-level spatial features from individual frames, while a recurrent neural network models temporal dependencies between consecutive frames to reveal inconsistencies characteristic of deepfake generation. The hybrid architecture is trained and evaluated on established benchmark collections, applying preprocessing steps that include face detection, alignment, normalization, and data augmentation. The trained model classifies input media as authentic or manipulated and supplies a confidence score. By combining frame-level feature extraction with sequence-level analysis, the system achieves competitive detection performance against diverse deepfake techniques and is suitable for deployment in social-media monitoring, forensic examination, and content-authentication platforms.
Disclaimer: Curated by HT Syndication.
