Benchmarking human face similarity using identical twins

The problem of distinguishing identical twins and non-twin look-alikes in automated facial recognition (FR) applications has become increasingly important with the widespread adoption biometrics. Due to high similarity both look-alikes, these face pairs represent hardest cases presented tools. This work presents an application one largest twin data sets compiled date address two FR challenges: (1) determining a baseline measure between (2) applying this determine impact doppelgangers, or on performance for large sets. is determined via deep convolutional neural network. network trained tailored verification task designed encourage group together highly similar embedding space achieves test AUC 0.9799. proposed provides quantitative score any given faces been applied large-scale identify pairs. An additional analysis that correlates comparison returned by tool also performed.