Accurate and Time-saving Deepfake Detection in Multi-face Scenarios Using Combined Features

Abstract

There has been an increasing interest in Deepfake detection because of the hidden risks that Deepfake technology poses for social privacy and security. Nowadays, many models achieve impressive performance on existing public benchmarks. However, the majority of existing methods are restricted to single-face scenarios. In this paper, we propose a model that can perform accurate and time-saving Deepfake detection in multi-face scenarios. We fuse different levels of features to improve the performance of the model and use single-face data to aid the training of the multi-face data. Our apporach achieves the state-of-the-art performance in multi-face scenarios and comprehensible experiments have been conducted to demonstrate the soundness and validity of our model.