MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641048480 A) filed by Seshadri Rao Gudlavalleru Engineering College; Appinedi Venkata Poorna Rajesh; Shaik Shakira; Shaik Sharhana; Vinnakota Amani; and Valluru Bhargav Kumar, Gudlavalleru, Andhra Pradesh, on April 16, for 'vision transformer-based deep learning framework for image and video deepfake detection.'
Inventor(s) include Seshadri Rao Gudlavalleru Engineering College; Appinedi Venkata Poorna Rajesh; Shaik Shakira; Shaik Sharhana; Vinnakota Amani; and Valluru Bhargav Kumar.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The rapid advancement of Artificial Intelligence has enabled the creation of highly realistic deepfake images and videos, raising serious concerns about digital security, privacy, and the spread of misinformation. In this project, a deepfake detection system based on the Vision Transformer (ViT) architecture is proposed to accurately identify manipulated visual content. Unlike traditional Convolutional Neural Networks (CNNs), the proposed model uses self-attention mechanisms to capture both local and global features from images and video frames, which helps in detecting even subtle manipulation patterns that are difficult to identify using conventional methods. The system processes input data through multiple stages, including frame extraction from videos, data preprocessing, feature extraction, and classification, to effectively distinguish between real and fake content. The model is trained on a diverse dataset containing both real and deepfake samples, ensuring robustness across different conditions such as lighting variations, facial expressions, and video quality. Various preprocessing and data augmentation techniques are applied to improve model performance and generalization. The system is evaluated using standard metrics such as accuracy, precision, recall, and F1-score, demonstrating reliable and consistent results. To enhance usability, the model is deployed using a FastAPI backend along with a web-based interface, allowing users to easily upload images or videos and receive real-time predictions within a few seconds. This makes the system practical for real-world applications. Overall, the proposed framework provides an efficient, scalable, and user-friendly solution for deepfake detection. It can be widely used in areas such as digital forensics, social media content verification, identity protection, and cybersecurity. In the future, the system can be further improved by incorporating advanced transformer models, multimodal data analysis, and real-time video processing capabilities to handle more complex and evolving deepfake techniques."
Disclaimer: Curated by HT Syndication.
