Qualitative Organization of Photo Collections via Quartet Analysis and Active Learning

We introduce the use of qualitative analysis and active learning to photo album construction. Given a heterogeneous collection of photos, we organize them into a hierarchical categorization tree (C-tree) based on qualitative analysis using quartets instead of relying on conventional, quantitative image similarity metrics. The main motivation is that in a heterogeneous collection, quantitative distances may become unreliable between dissimilar data and there is unlikely a single metric that is well applicable to all data. Our qualitative analysis utilizes multiple distance measures and applies them where reliable comparisons are possible. Then from the C-tree, we develop an active learning scheme for fine-grained photo scene classification, allowing the selection of representative photos for layout construction which better reflects user intent. Finally, the selected photos are laid out in a comic-like arrangement based on a style template library and layout optimization. Experiments demonstrate that our method is efficient, user-centered, and produces photo albums that are more preferable in comparison with previous approaches.