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

Table 6

Classification result after proposed optimization algorithm (experiment 4).

Classifier	Recall (%)	Precision rate (%)	F1 score (%)	FNR (%)	Accuracy (%)	Time (sec)
Linear SVM	99.4	99.94	99.94	0.6	99.9	108.64
Quadratic SVM	99.98	99.96	99.96	0.02	100	142.31
Weight KNN	99.88	99.9	99.89	0.12	99.9	179.7
Cosine KNN	99.86	99.86	99.86	0.14	99.9	51.49
Linear discriminant	99.68	99.7	99.69	0.32	99.7	53.094
Medium neural network	99.74	99.74	99.74	0.26	99.9	63.839
Narrow neural network	99.9	99.7	99.81	0.1	99.9	71.045
Wide neural network	99.92	99.96	99.94	0.08	99.9	43.207
Bilayered neural network	99.94	99.92	99.93	0.06	99.9	86.704
Trilayered neural network	99.94	99.92	99.93	0.06	99.9	97.371