Face-changing video detection method based on long-term and short-term memory network

Abstract

The invention discloses a face-changing video detection method based on a long short-term memory network. The detection method comprises the steps of video frame extraction, image feature extraction and long short-term memory network training test. The video frame extraction is responsible for extracting a key frame in a video clip and continuous multi-frame images after the key frame, cutting a face area in the image, processing the face image by using high-pass filtering, and extracting detail information in the face image; performing feature extraction on an image by using an Xception convolutional neural network trained in an Image Net image classification data set; and taking the output of the convolutional neural network as the features of the images, splicing the features extractedfrom each frame of image into a feature sequence, and inputting the feature sequence into a long short-term memory network for training to finally obtain a high-precision face-changing video classifier. According to the method, the inter-frame inconsistency existing in the forged video is fully utilized, the detection precision of the forged video is greatly improved, and a very good classification effect is achieved.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition, and in particular relates to a face-changing video detection method based on a long-short-term memory network. Background technique [0002] At present, video, as an important content of communication, plays an important role in people's life. At the same time, video can also be used as evidence in court. However, in recent years, due to the development of technologies such as deep learning, forgery technology for videos has reached a very high level. Among them, the deep forgery technology can generate a fake face video by inserting a synthetic face into the original video. On the one hand, this type of face-changing video is widely used in pornographic videos, which to a large extent causes great damage to the image of the person whose face is changed; on the other hand, this type of face-changing video can cause people to misjudge and even affect to people's decisions. Deepfake...

AI Technical Summary

Problems solved by technology

Existing detection methods mainly use the image defects caused by the face-changing process for detection. These defects mainly include problems such as synthetic edge chromatic aberration and resolution inconsistency caused by embedding the synthetic face into the original video.

However, video artifacts are easily masked in lower-resolution videos, making such detection methods much less effective

[0004] The second type is the detection method based on the consistency between frames. This type of scheme is mainly based on the consistency between the frames of the video. In the original face video, there should be a certain correlation between adjacent frames. The range of change is very small, but in the fake face video, since each frame of image is made separately, there is no time correlation, so there will be a sudden change in expression

However, this solution can easily cause problems such as network non-convergence when training recurrent neural networks, and the calculation cost is high.

Images