Abstract

Most saliency estimation methods aim to explicitly modellow-level conspicuity cues such as edges or blobs and mayadditionally incorporate top-down cues using face or textdetection. Data-driven methods for training saliency models using eye-fixation data are increasingly popular, particularly with the introduction of large-scale datasets anddeep architectures. However, current methods in this lat-ter paradigm use loss functions designed for classification or regression tasks whereas saliency estimation is evaluated on topographical maps. In this work, we introduce a new saliency map model which formulates a map as a generalized Bernoulli distribution. We then train a deep architecture to predict such maps using novel loss functions which pair the softmax activation function with measures designed to compute distances between probability distributions. We show in extensive experiments the effectiveness of such loss functions over standard ones on four public benchmark datasets, and demonstrate improved performance over state-of-the-art saliency methods.