Trust Stamp launches biometric facial image quality assessment API
Trust Stamp has published a white paper playing up its facial attribute and image quality assessment API, which the vendor says and be integrated with “any” relevant facial recognition, presentation attack detection or age verification apps.
The tools are designed to make provide feedback when capturing facial images to systems (or, separately, people if the vendor’s work on large language models pay off). Better automated image capture should make for better biometric enrollments and re-authentications, according to the document, which spends more time on face algorithms and less time describing image-capture quality.
The software tools incorporate and modify portions of ISO/IEC 29794-5, but they also introduce new recommendations based on previous standardization efforts. A new international standard for facial image quality is currently under development by an ISO/IEC working group.
Focusing on face metrics in individual facial images can be computationally simpler than categorizing images by image quality – blurriness, for example, according to the white paper.
Trust Stamp claims in the white paper shared with Biometric Update that its software tools pull subject data from deep learning-based facial attribute classifiers, bringing to the surface the presence of facial hair, for instance. The company refers to the suite of tools as “FAIQA” (for “facial attribute and image quality assessment”).
The company sees its quality assessment tools addressing a number of use cases, including video-frame selection for both facial recognition and proof of liveness. Quality summarization can spotlight devices and systems that are underperforming.
They can also cull poor images to improve biometric accuracy.