Year: 2025
Type: article
Abstract: The rise of AI-based tools has made it easier to create highly realistic deepfake videos, which pose significant risks such as political manipulation, blackmail, and the fabrication of terrorism-related content. These manipulated videos are often difficult to detect due to their minimal traces of alteration. In this project, we propose a Local Binary Pattern-based Convolutional Neural Network (LBPNET) for deepfake detection. The approach involves extracting Local Binary Pattern (LBP) features from facial images and training a Convolutional Neural Network (CNN) on these descriptors to develop a robust classification model. When a new test image is uploaded, the trained model evaluates it to determine whether it is real or fake.This method enhances the reliability of deepfake detection by leveraging texture-based feature extraction and deep learning. The following sections provide further details on the LBP technique and its role in our detection framework.
Author: Y. Ravindra Nadh
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4408911453