TABLE 6.

The classification accuracy rate based on high-level fusion.

Feature type	Rice cultivar	Model	Tr (%)	Val (%)	Te (%)	Model	Tr (%)	Val (%)	Te (%)	Model	Tr (%)	Val (%)	Te (%)
Full	Zhufujing83	SVM	100	100	90.38	LR	100	100	88.46	CNN	100	100	100
	AD516		100	100	93.75		100	100	98.44		100	100	87.50
PCA features	Zhufujing83		100	100	98.08		100	100	98.08		100	100	96.15
	AD516		100	100	96.88		100	100	98.44		100	100	90.63
AE features	Zhufujing83		100	100	80.77		99.17	90.00	75.00		100	100	73.08
	AD 516		100	85.00	90.63		100	95.00	89.06		100	90.00	85.94

Tr, training set; Val, validation set; Te, test set; Full, Full spectra; PCA features, featuers extracted by principal component analysis; AE features, features extracted by autoencoder.