Table 5.

Validation statistics of different models for fresh weight (g/m²) estimation.

Sensor	Feature type	Variables num	Metrics	RR	SVM	RF	GPR	KNN	Stacking	BMA
RGB	Spe	16	R 2	0.639	0.626	0.646	0.624	0.639	0.663	0.665
			RMSE	1782.5	1828.8	1754.4	1805.6	1772.2	1711.6	1704.8
	Str	2	R 2	0.859	0.818	0.849	0.831	0.846	0.859	0.861
			RMSE	1100.4	1266.0	1140.1	1232.4	1147.7	1103.1	1095.0
	Tex	9	R 2	0.787	0.761	0.764	0.759	0.775	0.803	0.803
			RMSE	1364.1	1467.8	1434.8	1461.4	1406.0	1311.7	1307.5
	Spe + Str	18	R 2	0.866	0.784	0.851	0.786	0.799	0.87	0.873
			RMSE	1077.0	1376.1	1133.6	1376.8	1323.9	1063.7	1046.7
	Spe + Tex	25	R 2	0.766	0.743	0.761	0.749	0.778	0.797	0.799
			RMSE	1437.5	1513.6	1446.0	1491.1	1395.9	1332.1	1323.1
	Str + Tex	11	R 2	0.866	0.805	0.846	0.805	0.804	0.865	0.868
			RMSE	1079.9	1313.9	1151.0	1321.0	1304.7	1084.8	1065.9
	Spe + Str + Tex	27	R 2	0.871	0.778	0.849	0.784	0.8	0.877	0.879
			RMSE	1058.5	1394.2	1140.0	1386.2	1320.9	1035.8	1022.4
MS	Spe	22	R 2	0.857	0.849	0.856	0.846	0.838	0.865	0.865
			RMSE	1121.8	1176.0	1124.7	1173.4	1203.2	1088.7	1088.5
RGB + MS	Spe + Str + Tex	49	R 2	0.858	0.793	0.876	0.799	0.823	0.888	0.887
			RMSE	1118.6	1348.0	1035.9	1332.3	1242.2	988.9	987.3

Note: Spe: spectral features; Str: structure features; Tex: texture features.