Table 2 Mean accuracies of the different cultivar classification models using leaves from each period in the soybean dataset independently

From: MFCIS: an automatic leaf-based identification pipeline for plant cultivars using deep learning and persistent homology

Method

Accuracy (%)

R1

R3

R4

R5

R6

IDSC + DP

19.67

23.67

20.33

20.00

16.00

HSC

27.02

31.07

31.20

30.60

27.71

PH

16.70

17.15

18.20

17.25

12.70

DF-VGG16/LDA

18.87

22.50

19.90

15.90

13.37

Fine-tuned Xception

30.60

35.97

33.37

29.73

20.40

MFCIS (Our model)

55.90*

61.40*

61.37*

59.80*

44.87*

  1. All the models were tested with ten different train-and-test splits. Only the Top-1 accuracy is listed. The accuracies in bold are the results of the proposed method. The superscript “*” denotes a significant difference (P < 0.05) between the proposed method and the other models using one-way ANOVA.
Back to article page