Saliency Maps on Image Hierarchies

Saliency maps for salient object segmentation created using two models based on a hierarchical image segmentation, a tree-like structure that represents regions at different scales from the details to the whole image (e.g. gPb-UCM, BPT).

The first model (HP) is based on a hierarchy of image partitions. The saliency at each level is computed on a region basis, taking into account the contrast between regions. The maps obtained for the different partitions are then integrated into a final saliency map by hierarchical inference.

The second model (SOH) directly works on the structure created by the segmentation algorithm, computing saliency at each node and integrating these cues in a straightforward manner into a single saliency map.

Results on ASD, MSRA, ECSSD and PASCAL1500 datasets.