Improved U-Net-Like Network for Visual Saliency Detection Based on Pyramid Feature Attention

Table 3

MAE values for the proposed model and three comparison FCN-based models on five test datasets. The results for RFCN, DCL, and UCF are from literature [35]. A smaller MAE value indicates a better performance, and the best result is highlighted in bold.

	ECSSD	DUT-OMRON	PASCAL-S	HKU-IS	DUTS-TE	Average
RFCN	0.109	0.111	0.133	0.089	0.090	0.106
DCL	0.151	0.157	0.181	0.136	0.149	0.155
UCF	0.080	0.132	0.127	0.074	0.117	0.106
Ours	0.089	0.095	0.115	0.076	0.080	0.091