Analyzing Human Observer Ability in Morphing Attack Detection—Where Do We Stand?

Morphing attacks are based on the technique of digitally fusing two (or more) face images into one, with final visage resembling contributing faces. Morphed not only pose a challenge to Face-Recognition Systems (FRS) but also experienced human observers due high quality, postprocessing eliminating any visible artifacts, and further printing scanning process. Few studies have concentrated examining how can recognize morphing attacks, even as several publications examined susceptibility automated FRS offered attack detection (MAD) approaches. MAD approaches base their decisions either single image no reference compare against (Single-Image (S-MAD)) or using (Differential (D-MAD)). One prevalent misconception is that an examiner’s observer’s capacity for facial morph depends subject expertise, experience, familiarity issue. No works reported specific results who regularly verify identity (ID) documents jobs. As involved in checking ID having images, lapse competence result significant societal challenges. To assess observers’ proficiency, this research first builds new benchmark database realistic from 48 different subjects, resulting 400 morphed images. Unlike previous works, we capture Automated Border Control (ABC) gates mimic border-crossing scenarios D-MAD setting probe study ability detect A dataset 180 produced S-MAD environment. In addition creating evaluation platform conduct analysis, employs 469 410 primarily governmental employees more than 40 countries, along 103 control group members examiners. The analysis offers intriguing insights highlights lack expertise failure sizable number by observers. Human tend lower accuracy compared four algorithms evaluated work. intended aid development training programs will prevent security failures while determining whether bona fide altered.