Page 179 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

168 Chapter 6 Plant leaf disease classification based on feature selection

Table 6.2 Confusion matrix of AlexNet model.

Class C1 C2 C3 C4
C1 37 0 0 0
C2 7 4 0 1
C3 0 0 27 0
C4 5 1 0 3
C1, anthracnose; C2, gall midge; C3, healthy; C4, powdery mildew.

4.1 Conventional models
VGG16 achieved 74.63% of training accuracy and 83.52% of ac-
curacy on the test set. AlexNet achieved 74.34% of training accu-
racy and 66% accuracy on the test set. As expected, with more
convolutional layers, VGG model shows better ability to remove

Figure 6.12 Training and validation result of VGG.