- 期刊
Deepfake Detection Based on No-Reference Image Quality Assessment (NR-IQA)
摘要
The deepfake technique has been shown its capability of replacing the actor's face in movie episodes or imitating the voice of an actor whose vocal cord is damaged. Due to the rapid development of digital technology, the use of deepfake images or videos has the capability of deceiving human eyes and computer programs with complex algorithms. However, malicious uses of this technology will seriously endanger national security and social stability. Recently, due to the widespread use of multimedia on the Internet, no-reference image quality assessment (NR-IQA) researches are popular. NR-IQA techniques are the image quality assessment methods without the reference images. That is useful for evaluating the image and video quality in social media. In this paper, we propose a novel deepfake detection method based on NR-IQA techniques. According to the experimental results, the detection rate of the proposed method is about 66.67%. Besides, we find out the resolution of deepfake videos should be another important factor to be considered.