A Feature Set for Neural Texture Video Detection Based on Scenes and Textures

The presence of fabricated videos poses a significant challenge in the present era of social media dominance. Various categorizations for counterfeit videos exist, with Neural Textures being a prominent example. The task of identifying such fraudulent videos is intricate. This research endeavour aims to grasp the distinct attributes associated with Neural Texture videos. In this pursuit of comprehending Neural Texture videos, this study delves into the distinguishing traits that define them. Consequently, the study employs scene and texture detection techniques to formulate a distinctive set of 19 data features. This feature set is designed to discern whether a video exhibits Neural Texture characteristics or not. To validate this set, a standard dataset of video attributes is employed. These attributes are subjected to analysis using a machine learning classification model. The outcomes of these experiments are evaluated through four distinct methodologies. The assessment discloses favourable performance outcomes when employing the machine learning approach and the proposed feature set. Based on these findings, it can be inferred that utilizing the suggested feature set enables the prediction of whether a video displays Neural Texture characteristics or not. This thereby confirms the hypothesis that a correlation exists between a video's attributes and its authenticity, specifically in determining whether the video qualifies as a Neural Texture.