What to look for in a liveness detection solution

Facial liveness detection technology has become more commonplace in recent years as a means to combat identity fraud, and it is important to understand the different aspects of this technology when considering a vendor. In this article, we will discuss five important points to consider when evaluating liveness detection technology vendors. We will discuss how they score in third-party presentation attack detection tests, the level of bias in their facial recognition, whether they offer passive or active liveness detection, integration requirements, and privacy considerations.

• How do they score in 3rd party presentation attack detection PAD tests? Check who has accredited them and what level they’ve achieved. ISO/IEC 30107-3 has 3 levels (1, 2, 3), as does FIDO (A, B, C). The lower levels cover more simple attacks, such as printed faces, or faces on a screen. While the higher levels cover more sophisticated attacks, such as attacks involving specialist masks, 3D printed faces, or deepfakes.
• How biased is their facial recognition? Facial recognition in general has biases towards darker-skinned people and women. Ask how their solution performs across diverse groups and what the company is doing to lower AI-bias in the industry.
• Do they offer passive or active liveness detection, or both? Ask for a breakdown of the differences for your use case. Passive liveness detection typically has a lower drop off than active liveness detection. If the vendor delivers both, then ask about the difference in drop-off rates during the initial and ongoing authentication process.
• How well does it integrate into your existing authentication setup? Ensure it has the right API and/or SDK integration for your needs.
• How do they respect your users privacy and adhere to privacy regulation? Some liveness detection solutions store patterns in lieu of your users liveness selfies. If they do store photos, look out for encryption at rest, and encryption in transit.

Facial liveness detection technology is a valuable tool for combating identity fraud. When considering a vendor, it is important to understand their approach to presentation attack detection tests, the level of bias in their facial recognition technology, their liveness detection capabilities, the level of integration required, and the privacy considerations. By understanding these points and evaluating vendors on these metrics, you can ensure that you are selecting the best solution for your needs.