Anomaly detection, which is approaching the problem of face anti-spoofing as a one-class classification problem, is emerging as an increasingly popular alternative to the traditional approach of training binary classifiers ...