anti-spoofing-of-rppg

the implement of 3D Mask Face Anti-spoofing with Remote Photoplethysmography

English|中文 )

Introduction of project

• config.py

  • Contains all the settings we need to configure

• rPPG_GUI.py

  • Entry function
  • ompleted the design of GUI interface
  • Detect if it is spoofing(Just a simple judgment, there is no data set, so SVM training is not used.)

• rPPG_preprocessing.py

  • Image preprocessing, face detection

• rPPG_processing_realtime.py

  • Extract rppg signals by fftlength's preprocessed photos

• rPPG_Extracter.py

  • Take a picture and detect the face, then count he RGB components of the face and find the average, finally cache them in self.rPPG

• FaceDetection.py

  • Face detection, we uses key point detection function of it

• Processing

  • Run the program (rPPG_GUI.py) to collect photos by camera or video collected , perform face detection on the photos (rPPG_preprocessing.py), and extract the average of RGB components (rPPG_Extracter.py). After preprocessing specified number (fftlength) photos, perform fft analysis and extraction rppg signal (rPPG_processing_realtime.py). Sliding window continuously extracts the signal and displays

• Run

  • Check if your camera is working by cheese command in linux bash
  • Configure config.py
  • run rPPG_GUI.py

• run rPPG_GUI.py

  • python3 rPPG_GUI.py

• reference

  • Remote Photoplethysmography Correspondence Feature for 3D Mask Face Presentation Attack Detection
  • Generalized face anti-spoofing by detecting pulse from face videos

• the implement of Face Anti-Spoofing Using Patch and Depth-Based CNNs

  • https://github.com/shicaiwei123/patch_based_cnn