Abstract:Current saliency detection methods mainly work on exploring the potential of low-level and high-level visual features, such as color, texture and face, but treat location information as a weak assistance or completely ignore it. In this paper, we reveal the importance of location information in saliency detection. We analyze the largest public image dataset for saliency detection THUS10000, and and the relationship between content location and saliency distribution. To further validate the effect of location information, we propose two location based saliency detection approaches, location based Gaussian distribution and location based saliency propagation, which make use of no or weak assistance of image content. Experimental results show that location based saliency detection can obtain much better performance than random selection, even better than most state-of-the-art saliency detection methods.

How Important is Location in Saliency Detection