Two-Stage Intelligent DarkNet-SqueezeNet Architecture-Based Framework for Multiclass Rice Grain Variety Identification

Table 3

Classification results of the DarkNet19 model (experiment 1).

Classifier (%)	Recall (%)	Precision (%)	F1 score (%)	FNR (%)	Accuracy (%)	Time (sec)
Linear SVM	98.28	98.3	98.29	1.72	98.3	277.71
Quadratic SVM	98.36	98.38	98.37	1.64	98.4	291.46
Weight KNN	98.22	98.2	98.21	1.78	98.2	1444.8
Cosine KNN	98.2	98.22	98.21	1.8	98.2	1781.5
Linear discriminant	98.26	98.28	98.27	1.74	98.3	73.55
Medium neural network	97.86	97.88	97.87	2.14	97.9	1203.3
Narrow neural network	97.76	97.74	97.75	2.24	97.8	1943.7
Wide neural network	97.9	97.88	97.89	2.1	97.9	1699.8
Bilayered neural network	97.82	97.84	97.83	2.18	97.8	2330.2
Trilayered neural network	97.82	97.88	97.85	2.18	97.8	1922.5