Attention Based Album Slideshow

Slideshow is a popular way to display a digital album automatically. However, it is time consuming and boring when the album is large. Actually, while browsing an album manually, an audience may spend different length of time on different photos. Considering this, in this paper, we propose an approach to estimate the attention amount attracted by each photo and further predict its duration. Three types of attention related factors are considered. 1) The subjective factor, described through photo content. 2) The objective factor, which means the photographic quality of a photo. 3) The album context, represented through a photo´s novelty in the album. Then Support Vector Regression is employed to estimate the duration. The proposed slideshow method displays the photos according to the estimated duration. Both objective and subjective evaluations show the effectiveness of the proposed method.