Deeper and Mixed Supervision for Salient Object Detection in Automated Surface Inspection

Table 5

SDMS network’s performance on RSDDs, road cracks, and NEU datasets. We used the same model and hyperparameter settings in the training.

Dataset		RSDDs		Road cracks		NEU		Settings
Model								The same as the SDMS-D, which is shown in Table 3
SDMS	Output0	0.01132	0.5526	0.03044	0.6973	0.00891	0.7927
	Output1	0.00549	0.7842	0.01955	0.8169	0.00616	0.8229
	Output2	0.00527	0.7835	0.01934	0.8148	0.00593	0.8249
	Output0+	0.00906	0.6884	0.02728	0.6549	0.00735	0.7913
	Output1+	0.00602	0.7888	0.01494	0.7805	0.00611	0.8258
	Output2+	0.00575	0.7882	0.01479	0.7786	0.00590	0.8306