Saliency Detection Using Sparse and Nonlinear Feature Representation

Table 3

Comparison of sAUC score of the proposed model with 13 other state-of-the-art models.

Dataset	AIM [16]	GBVS [37]	MESR [39]	ICL [46]	Itti [37]	MPQFT [39]	SDSR [26]	SUN [47]	HFT [2]	LG [29]	ERDM [17]	ΔQDCT [39]	AWS [23]	Proposed
TORONTO [16]	0.699	0.6404	0.7141	0.6476	0.6481	0.7119	0.7062	0.6661	0.6915	0.6939	0.7023	0.7172	0.7130	0.7179
Optimal σ	0.04	0.01	0.05	0.03	0.01	0.04	0.03	0.03	0.06	0.04	0.04	0.05	0.01	0.04
KOOTSTRA [34]	0.5901	0.5572	0.5978	0.5769	0.5689	0.5985	0.5984	0.5613	0.5911	0.5938	0.603	0.5998	0.6185	0.6157
Optimal σ	0.03	0.01	0.04	0.01	0.04	0.05	0.04	0.02	0.05	0.01	0.03	0.03	0.01	0.03
IMSAL [2]	0.7424	* 0.7665	0.7153	0.7364	0.7512	0.71694	0.7403	0.6955	0.7419	*0.7356	0.7455	*0.7434	0.7468	0.7560
Optimal σ	0.08	0.07	0.15	0.13	0.09	0.14	0.13	0.16	0.09	0.12	0.12	0.10	0.05	0.09

The top ranked model is in bold font and second ranked model is in italic font. Here results with optimal average sAUC of each method with the corresponding Gaussian blur(σ) are reported. As in [47], we repeat the sAUC calculation for 20 times and compute the standard deviation of each average sAUC, which ranges from 1𝐸−4 to 5𝐸−4. (*values taken from [48]).