Abstract: The present disclosure relates to systems, methods, and non-transitory computer readable media for selecting representative recent cover images from collections of digital portraits by determining selection scores based on average face vectors. For example, the disclosed systems can generate an average face feature vector to represent a common appearance or facial expression of a user in the collection of digital portraits. The disclosed systems can further determine representativeness scores that indicate measures of closeness of digital portraits to the average face feature vectors. In addition, the digital portrait selection system can determine various other factors, such as recency scores, face area scores, and face expandability scores. Based on these factors, the digital portrait selection system can determine an overall selection score and select a digital portrait as a cover image.

Abstract: The present disclosure relates to systems, methods, and non-transitory computer readable media for selecting representative recent cover images from collections of digital portraits by determining selection scores based on average face vectors. For example, the disclosed systems can generate an average face feature vector to represent a common appearance or facial expression of a user in the collection of digital portraits. The disclosed systems can further determine representativeness scores that indicate measures of closeness of digital portraits to the average face feature vectors. In addition, the digital portrait selection system can determine various other factors, such as recency scores, face area scores, and face expandability scores. Based on these factors, the digital portrait selection system can determine an overall selection score and select a digital portrait as a cover image.