Machine Learning Based Datasets Collection and Preprocessing of Deepfake Video

Abstract

In a matter of developing deepfake technology is a major challenge in methods to detect manipulated videos. This study is dedicated to the deepfake dataset review and discusses the possible strategies in deepfake Collecting and formatting data for the establishment of a reliable deepfake detection model. We survey different datasets available for deepfake research and refer to the preprocessing techniques that aid in the performance of deepfake detection models and provide an exhaustive account of the existing deepfake video datasets. The study suggests how can selecting the right data sets and methodologies for preprocessing in order to increase the accuracy and efficacy of deepfake. present the issues and limitations of current datasets and preprocessing methods and envisage future work such as the creation of novel datasets and sophisticated preprocessing methods.