Segmentation and counting of wheat spike grains based on deep learning and textural feature

Table 5 Relationships between counting with methods I and II and manual counting of 5 varieties

Wheat varieties	Counting accuracy of method I				Counting accuracy of method II
Wheat varieties	Mean absolute error	Mean relative error/%	RMSE	R²	Mean absolute error	Mean relative error/%	RMSE	R²
Bainong 4199	2.90	2.78	3.27	0.83	2.10	2.56	2.51	0.90
Kexing 3302	1.60	2.41	2.14	0.85	1.40	2.32	1.67	0.91
Yangmai 15	3.60	3.04	4.56	0.81	1.90	2.56	2.02	0.92
Yunong 904	4.10	3.28	4.44	0.82	2.60	2.81	3.52	0.89
Zhengmai 136	1.60	2.37	2.14	0.86	1.90	2.47	2.88	0.87
All cultivars	3.74	4.23	3.74	0.81	1.98	2.75	2.60	0.91

ISSN: 1746-4811