Determination of Wheat Heading Stage Using Convolutional Neural Networks on Multispectral UAV Imaging Data

Table 2

Fitting effect of traditional VI and heading rate.

Index	Method	Fitting result
		R^2	RMSE
DI	NN	0.23	0.15
	SVM	0.40	0.20
	DT	0.55	0.31
	1D CNN + DT	0.57	0.31
DVI	NN	0.26	0.16
	SVM	0.28	0.17
	DT	0.63	0.35
	1D CNN + DT	0.70	0.36
NDVI	NN	0.27	0.17
	SVM	0.44	0.22
	DT	0.64	0.39
	1D CNN + DT	0.66	0.39
CIrededge	NN	0.23	0.15
	SVM	0.76	0.32
	DT	0.72	0.36
	1D CNN + DT	0.74	0.37
GNDVI	NN	0.27	0.17
	SVM	0.42	0.21
	DT	0.60	0.36
	1D CNN + DT	0.62	0.36
TVI	NN	0.13	0.31
	SVM	0.78	0.29
	DT	0.85	0.41
	1D CNN + DT	0.87	0.42