Advanced deep learning methods and a new face finder are contributing to substantial performance upgrades for Cognitec’s FaceVACS face recognition engine. In particular, the new version shows higher accuracy rates for facial images with masks, strong poses and extreme mouth expressions. Extensive research and development efforts to enhance the company’s face finder technology resulted in a tenfold higher detection rate for faces largely covered by facial masks.
A specialised algorithm version with less memory consumption serves Linux-on-Arm and Android platforms. For this algorithm, test results showed particularly high matching accuracy for images of people wearing face masks. Cognitec is already preparing the release of the next matching algorithm, expecting a further accuracy boost for general matching tasks, and especially for comparing images with partially occluded faces.
