Spoofing BioID Face Liveness Detection 2024 | 3D Silicon Mask | Anti-Spoofing Deepfake Detection

In this biometric #spoofing challenge, BioID’s anti-spoofing #livenessdetection was tested to demonstrate its effectiveness against presentation attacks. The system was attacked with photo and video replay attacks, presented on various devices. In addition, it had to cope with fraud attempts like silicon masks, plastic and T-Shirt masks as well as individually manufactured 3D paper masks. With a standard Microsoft Webcam, BioID’s Liveness Detection demonstrator analyses in real-time if a presentation attack is happening. To visualize the results, it surrounds the faces either in a green frame (live person) or in a red one (fake).

What is Liveness Detection in general?

Liveness Detection describes a biometric system that is able to differentiate between a live person and a spoof/fake. The system therefore detects if the person looking into the camera is really present in the moment of the image or video capture, or instead, a spoof artifact was used. In order to perform this operation, it then uses algorithms that analyze the data, collected with any standard smartphone or webcam. The ultimate goal hereby is to mitigate the vulnerability of biometric systems to spoofing attacks. No biometric scanners or special readers are required for software-based systems to prevent fraud through biometric imitations.

How does BioID’s ISO compliant liveness detection work?

First, we collect two images from the person looking into the camera of the electronic device, only to then check for natural motion and facial features. Here, our very own algorithm can tell the difference between a real 3D face and a 2D photo due to their different motion features. Also, we use Artificial Intelligence through powerful DCNNs (deep convolutional neural networks) enabling the system to detect different, more sophisticated spoofing and presentation attacks.

BioID's liveness detection is compliant to the international standard for presentation attack detection, ISO/EIC 30107-3 as audited by the FIDO accredited biometric testing laboratory TÜViT from Germany. In addition, the PAD (presentation attack detection) has been part of a customer's full FIDO biometric component certification and tested by the French laboratories Elitt and Leti for compliance.

What Spoofing attacks are there?

Generally, the most common spoofing occurs as a presentation attack on the sensor-level of the electronic device. Like shown in the video, there is a variety of different presentation attacks:

o Photos and tablet videos
o Flat paper, resin, latex or silicone masks
o 3D digital doubles
o Avatars and #deepfakes presented on screens
o 3D prints, wax heads or sculptures

What other methods are there to detect spoofing?

Besides software-based methods, which can be used with any type of camera, some systems also rely on special hardware. Those are mostly used in controlled environments like airports or border control. Such high-end expensive sensors include 3D and infrared cameras and are not applicable to end-user scenarios like fraud prevention during digital identity verification.

In what scenarios is Liveness Detection usable or even necessary?

Nowadays, anti-spoofing methods like Liveness Detection are of use in multiple different areas, contributing to a safer digital presence. Especially in scenarios where private data could be exploited and identity theft is possible, ensuring an effective solution for verification processes is necessary. Its use cases span from home office authentication to as far as health care scenarios. On top of that, it is also relevant every time some form of identity verification is needed where providing an ID online could lead to data breaches and exploitation.

What impact did the pandemic have on the biometric industry?

Especially since the Covid-19 outbreak, liveness detection systems have become crucial in remote identification processes as individuals and employees largely act from home and in unsupervised situations.

Final question: Can I try BioID’s Liveness Detection or other products?

Sure, you can. Just head on over to our playground of our website to try our patented software.

For more information about BioID you can also subscribe to:

[iliade_edaili]

The other day I checked with the face of a "baby reborn" doll.
And latex masks, made up and with skin textures, pass easily through your system.

[bioidgmbh]

Your system can be spoofed! That's what we hear when someone injected a video into a system as a virtual camera attack. Well, not really! The spoofing target in that attack scenario was the application, and NOT our technology (our API and the algorithms performing the biometrics were not even touched). There is a major difference between application level attacks (need to be countered by a secure app, preventing any virtual camera access, for instance) and a presentation attack. The latter is exactly what BioID's liveness detection is aimed at and what is demonstrated in this video: It prevents any presentation attacks (fake biometrics presented to the camera), such as video replays, deepfakes presented on screens, 3D paper, silicon and other masks, etc. There is no question that an application level attack needs to be prevented, as well, to make identity verification secure. But, BioID does not build end-user applications, and instead supports their customers with reliable, ISO 30107-3 compliant PAD/liveness detection. When working with our customers we of course support and consult them concerning the construction of a secure app. Together, we join forces with our partners to prevent criminals from harming individuals' identities.