Deepfake Video Detection System Using Deep Neural Networks

Abstract

A deepfake is a type of synthetic media that uses artificial intelligence and machine learning techniques to manipulate or superimpose images, videos, or audio onto existing footage in a way that appears authentic and realistic, often with the intent to deceive or mislead viewers. There are several approaches to using neural networks for deep fake detection. One approach is to use a convolutional neural network (CNN) [1] to analyze the visual artifacts in the image or video. The CNN can detect inconsistencies or anomalies in the image or video that are indicative of manipulation, such as differences in lighting or blurring at the edges of the image. ResN et-50 has been used in deepfake detection by training the network on a large dataset of real and fake videos In this paper, Resnet50 and LSTM [13] are combined to make a hybrid architecture are used in deep fake video detection as a web framework using python. Combining ResNet50 and LSTM can help to leverage the strengths of both architectures and improve the accuracy of deep fake video detection, especially for videos that involve both image-based and sequential data. A comparative analysis of different models were assessed using various datasets such as Celeb-DF, Face Forensic++ datasets.