/liveness-detection

A repository to compare and contrast different methods for liveness detection and trying an ensemble method in the end to avoid L1 and L2 spoofing stated as benchmark on liveness.com

liveness-detection

A repository to compare and contrast different methods for liveness detection and trying an ensemble method in the end to avoid L1 and L2 spoofing stated as benchmark on liveness.com

Literature and Inspiration

Adrian Rosenbrock's article on Liveness detection -

The approaches he mentioned are the ones which are used today mostly for this problem statement.

  1. Texture analysis - Computing Local Binary Patterns (LBPs) over face regions and using an SVM to classify the faces as real or spoofed.

  2. Frequency analysis - Examining the Fourier domain of the face.

  3. Variable focusing analysis- Examining the variation of pixel values between two consecutive frames.

  4. Heuristic-based algorithms - Eye movement, lip movement, and blink detection. These set of algorithms attempt to track eye movement and blinks to ensure the user is not holding up a photo of another person (since a photo will not blink or move its lips).

  5. Optical Flow algorithms - Examining the differences and properties of optical flow generated from 3D objects and 2D planes. 3D face shape, similar to what is used on Apple’s iPhone face recognition system, enabling the face recognition system to distinguish between real faces and printouts/photos/images of another person.

  6. Combinations of the above - Enabling a face recognition system engineer to pick and choose the liveness detections models appropriate for their particular application.

iBeta Liveness quality ISO Standards

Algorithms

  1. High res face masks with eyes and lips and nose cutouts

    Color Texture Analysis Paper Matlab Code

  2. Deep Learning CNN Approach -

    1. LivenessNet Pyimagesearch

    2. Model 3 proposed in this paper and source code on CASIA-FASD Dataset. The dataset has some restrictions of access. Try VPN (Southeast Asia) and registering here

  3. Red Eye + Texture Analysis

    This is a proposed approach which takes into consideration the hardware on which the face detection is to be run.

    Some references -

    1. Selfie Flash which gave me the idea that you don't need front flash in mobile device

    2. Red Eye Effect which takes away blinking plus video specific spoofing

    3. Face Texture found from the very interesting comparison paper for liveness detection

alt text