Hierarchical Multimodal Adaptive Fusion (HMAF) Network for Prediction of RGB-D Saliency

Table 1

Comparison of quantitative scores of different loss function on two datasets.

Datasets	Criteria	MSE	KL	CC′	MSE + CC′	MSE + KL	CC′ + KL
NUS	CC	0.5224	0.5363	0.5296	0.5533	0.5473	0.5452
	KLDiv	1.1690	1.2344	1.1834	1.0794	1.1907	1.1320
	AUC	0.8519	0.8637	0.8627	0.8422	0.8571	0.8613
	NSS	2.0926	2.1698	2.1259	2.3069	2.2391	2.2086
NCTU	CC	0.8256	0.8390	0.8287	0.8472	0.8345	0.8308
	KLDiv	0.3176	0.2736	0.4877	0.3228	0.3333	0.3438
	AUC	0.9025	0.9087	0.9011	0.9130	0.9044	0.9041
	NSS	2.2957	2.3137	2.3199	2.3493	2.3437	2.3111