Fast & Efficient Saliency

Feast & Efficient Saliency is a simple center-surround saliency detection model for still images. It is based on estimating saliency of local feature contrast in a Bayesian framework. The distributions needed are estimated particularly using sparse sampling and kernel density estimation. Furthermore, the nature of method implicitly considers what refereed to as center bias in literature. [paper]

Recently, we scored the method with MIT Benchmark. Our scores are AUC = 0.8033, Similarity = 0.4952 and Earth Mover’s distance = 3.3488. The parameters values for the above scores are as follows: α = 30, σ0 = 10, σ1 = 1, sampling radius is [13, 25, 38] and number of surround samples is 8. The overall score ranks us 3rd by 21st of April, 2013.