Supporting Human Examiners in Facial Image Manipulation Detection

Many countries lack live enrollment procedures for facial images in identity documents, allowing citizens to provide their own photos. This operational procedure opens avenues for fraudulent attacks, primarily through digital manipulations of facial images, which have an impact on automated identity verification performance. Detecting such manipulations before image inclusion into documents is critical, especially given the difficulty in identifying them visually. To address this, a software tool is proposed to assist human examiners in analyzing facial images for electronic ID documents. The tool aligns with guide-lines from the Facial Identification Scientific Working Group, implementing measures covering various facial components to ease the comparison between the ID photo and a live-captured image during enrollment.