Face detection is the first half of face recognition, and it consists of three consecutive steps, namely face detection, landmark detection, and face alignment. The process starts by locating where a face is or faces are in a visual input. Upon successfully detecting a face, the algorithm then goes on to extract landmark points that are to be used as it aligns and crops the image making it the right size so that it can be processed further.

Face matching is the latter half of face recognition, and made up of two consecutive steps including feature extraction and feature comparison. Upon receiving the aligned face image, the algorithm extracts a feature value by applying dozens of convolutional layers that gradually decrease the size of each component of a face. In the end, an nth dimension vector representing a unique face is created, and this information is then used for the verification of different people’s identity.

Face anti-spoofing is a function that detects whether the visual input received is from a real person physically present at the point of capture, and it runs not only on IR cameras, but also on RGB cameras. With a true positive rate as high as 99%, which then functions as a highly effective measure for preventing identity theft, ALCHERA’s anti-spoofing has been applied to a variety of industries where there is a need for such physical and/or digital security.

Mask detection is a function that is used to check whether a detected face is properly wearing a facial mask. With an accuracy of 99.70%, ALCHERA’s mask detection technology can tell whether a person is wearing a facial mask properly or not even when the person is wearing the mask below one’s nose or chin.

Age, Gender, Emotion(AGE) is a function that estimates the age, gender, and emotion evident in a person from a detected face. The most common usage of AGE is the collection and utilization of customer data for market research and/or targeted advertisement in retail businesses.