AN AUTOMATIC PERSONALIZED PHOTO RECOMMENDATION SYSYTEM BASED ON IMAGE VISUAL QUALITY AND USER PREFERENCES

Abstract

This paper presents an automatic personalized photo recommender system which recommends photos from a large collection. Our proposed system recommends photos based on userpreferences about aesthetics and visual quality features of the photo. A large dataset has been put together, which has been used to collect user-preferences. A random forest based learning system has been invoked to learn the user preferences about different image features including aesthetic features. The system is validated using a part of the collected user preferences as ground truth and it has been compared to a random selection of photographs. Our automatic system significantly outperforms the random selection, which shows the usefulness of our proposal, especially when the collection of photos is manually unmanageable.