Double-flow face counterfeiting detection method based on Swin Transform

A technology of forgery detection and dual-stream face, applied in neural learning methods, instruments, biological neural network models, etc., can solve problems such as insufficient generalization ability, poor anti-compression ability, weak generalization ability, etc., to improve generalization ability , resist compression, and enhance the effect of feature information

Pending Publication Date: 2022-07-01
SHANGHAI UNIV
View PDF0 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problems of insufficient generalization ability and poor anti-compression ability of existing face detection methods, and provides a dual-stream face forgery detection method based on Swin Transformer, which solves some limitations of existing face forgery detection schemes At the same time, the dual-stream framework improves the anti-compression ability of the model, making it more in line with the common face video quality in daily life.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Double-flow face counterfeiting detection method based on Swin Transform
  • Double-flow face counterfeiting detection method based on Swin Transform

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0033] see figure 1 , a dual-stream face forgery detection method based on Swin Transformer, including the following steps:

[0034] (1) Prepare the face video dataset:

[0035] Prepare the face forgery video dataset, and select the public face forgery video dataset: FaceForensics++;

[0036] (2) Crop the video into frames:

[0037] Considering that the training goal is to judge the authenticity of face images, the video data set is processed, specifically:

[0038] For all real or fake face videos, use the OpenCV tool library to divide them into non-overlapping video frames, and the specific method is to intercept at least 15 frames per second;

[0039] (3) Extract the face area:

[0040] Use the face detector to locate the face area, remove the surrounding irrelevant background information, extract the face area, and obtain the face area image dataset;

[0041] (4) Division of the dataset:

[0042] Use the face region image data set obtained in the previous step for tr...

Embodiment 2

[0049] This embodiment is basically the same as the first embodiment, and the special features are:

[0050] A dual-stream face forgery detection method based on Swin Transformer, comprising the following steps:

[0051] (1) Prepare the face video dataset:

[0052] First, we need to prepare the face forgery video dataset. The public face forgery video dataset is selected: FaceForensics++; the FaceForensics++ dataset is the first large-scale face forgery video dataset, which uses DeepFakes, FaceSwap, Face2Face, NeuralTextures to create different categories of face forgery techniques The first two techniques are facial identity replacement, and the latter two are facial expression reconstruction; this dataset also compresses the real and fake videos with three parameters of C0, C23, and C40 in H.264 encoding. It is to better simulate the quality of face videos spread on the Internet, in order to be closer to the face videos used in real life;

[0053] (2) Crop the video into ...

Embodiment 3

[0065] This embodiment is basically the same as the above-mentioned embodiment, and the special features are:

[0066] In this embodiment, a dual-stream face forgery detection method based on Swin Transformer, such as figure 1 shown, including the following steps:

[0067] First, the video data set needs to be preprocessed. The video data set here uses FaceForensics++. This data set contains four sub-data sets, namely DeepFakes and FaceSwap based on face swap forgery, and Face2Face based on facial expression reconstruction. And NeuralTexture, each video sub-dataset contains 1000 videos, plus 1000 real face videos, there are 5000 videos in total, which is a relatively large public video dataset. In order to simulate the video quality of real life scene propagation, this FaceForensics++ dataset also provides videos with three different compression qualities, namely C0 (no compression), C23 (light compression) and C40 (heavy compression).

[0068] The steps of video preprocessi...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention relates to a double-flow face forgery detection method based on Swin Transformer, which is used for detecting a face forgery image by using deep learning. A deep learning network model is integrally built, and the network model is divided into three parts: a double-flow network, a feature extraction network and a classifier. Because all face counterfeit data sets disclosed at present are videos, the videos need to be clipped into frame pictures by using OpenCV. In addition, the frame picture contains a large amount of background information, so that a human face area needs to be cut out by using a human face positioning algorithm. And inputting the obtained face region image into a double-flow network and a feature extraction network to extract and learn features. And finally, inputting the learned features into a classifier, and identifying whether the face image is true or false. The method is used for solving the problem of partial limitation, namely weak generalization ability, of the existing face counterfeiting detection scheme, and meanwhile, the compression resistance of the model is improved through the double-flow framework, so that the method is more in line with the common face video quality in daily life.

Description

technical field [0001] The invention relates to a dual-stream face forgery detection method based on SwinTransformer, which is applied to solve part of the limitation of the existing face forgery detection scheme - the problem of weak generalization ability, and also proposes a dual-stream framework to improve the certain anti-compression ability of the model. , making it more in line with the common face video quality in daily life. Background technique [0002] In the past few decades, with the rapid rise of the Internet, emerging social networks have been integrated into people's lives, making digital images and videos common digital data objects. Reportedly, nearly 2 billion pictures are circulated on the internet every day, with the huge use and dissemination of digital images, initially followed by the use of editing software like Photoshop to change the content of the images, but with computer vision With the latest progress in image processing technology and breakth...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06V40/16G06V20/40G06V10/30G06V10/774G06V10/764G06V10/82G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/24G06F18/214
Inventor 张新鹏潘照广冯国瑞
Owner SHANGHAI UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap