Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing age estimation. In one aspect, a method includes receiving an image of a person, submitting the image to multiple binary classifiers that are each trained to classify the person in the image as belonging to one of two predefined age groups, or as belonging or not belonging to a particular age group, where each output includes a confidence value associated with classifying the person in the image, obtaining the confidence values from the multiple binary classifiers, aggregating the confidence values, and generating an age estimation for the person in the image based on the aggregated confidence values.