Facial recognition technology is successfully used in our lives, both for private purposes and in more complex and serious situations, like border control, granting access, or as a working time controller. The system works accurately most of the time, granting extra security level, but it may be prone to false image identification due to improper pose, lighting, or resolution. To deal with quality assessment algorithms falsely rejecting or accepting images, FATE quality checks the ability of vendors to eliminate the issue. Who and how adequately and professionally can evaluate face analysis quality?
The role of NIST in the face analysis quality evaluation
When the evaluation of biometrics technologies is concerned, the evaluation of the tools is conducted by a range of prominent authorities, with the National Institute of Standards and Technology (NIST) at the forefront. The institute is part of the U.S. Department of Commerce and conducts a series of various evaluations, one of them called NIST FATE Quality.
The series of tests aims to estimate if image quality assessment algorithms can deal with face images of lower quality, taken in poor and challenging conditions. The testing includes the analysis of illumination, facial expressions, and focus. Visit Neurotechnology’s page to read more about NIST and the evaluations it performs.
What is FATE Quality Evaluation?
Facial recognition technology heavily relies on Quality Assessment Algorithms (QAA) to evaluate facial images based on factors such as pose, lighting, and resolution. However, these algorithms are not perfect and can lead to serious consequences if they inaccurately accept poor images or reject good ones.
Both situations can increase operational costs and compromise security measures. To address this problem, FATE Quality utilizes machine learning to evaluate QAA algorithms provided by vendors and improve their accuracy.
What is FATE MORPH evaluation?
It is crucial to verify the effectiveness of algorithms that detect face morphing attacks, in addition to evaluating their quality using FATE. The evaluation process, known as NIST FATE MORPH, identifies images that have been falsely created by combining multiple faces into a single image. The evaluation assesses various datasets and analyzes different morphing methods, including those that are widely known to the public and those that are only accessible to experts. See here an example of FATE.
