Amazon Rekognition
Developer Guide

Guidelines on Face Attributes

Amazon Rekognition returns a bounding box, attributes, emotions, landmarks, quality, and the pose for each face it detects. Each attribute or emotion has a value and a confidence score. For example, a certain face might be found as ‘Male’ with a confidence score of 90% or having a ‘Smile’ with a confidence score of 85%. We recommend using a threshold of 99% or more for use cases where the accuracy of classification could have any negative impact on the subjects of the images. The only exception is Age Range, where Amazon Rekognition estimates the lower and upper age for the person. In this case, the wider the age range, the lower the confidence for that prediction. As an approximation, you should use the mid-point of the age range to estimate a single value for the age of the detected face. (The actual age does not necessarily correspond to this number.)

One of the best uses of these attributes is generating aggregate statistics. For example, attributes, such as Smile, Pose, and Sharpness, may be used to select the ‘best profile picture’ automatically in a social media application. Another common use case is estimating demographics anonymously of a broad sample using the gender and age attributes (for example, at events or retail stores).