Table 1.
Results of previous work using deep learning for identification of plant diseases.
Fig 1.
The proposed palm tree disease identification using deep learning models.
Fig 2.
The proposed ResNet architecture.
Table 2.
The deep learning layers in each stack’s block in the proposed ResNet.
Table 3.
Another deep learning layers in each stack’s block in the proposed MResNet.
Fig 3.
Sample of palm leaves images.
Table 4.
The hyper parameters used in training options for all deep learning ResNet models.
Fig 4.
Training progress of the proposed ResNet 1x1, 3x3 with and without data augmentation.
(a) With data augmentation; (b) Without data augmentation.
Fig 5.
Palm-Leaves classification result using the proposed ResNet 1x1, 3x3 with and without data augmentation.
(a) With data augmentation; (b) Without data augmentation.
Fig 6.
The confusion matrix of the classification accuracy using the proposed ResNet 1x1 and 3x3 with and without data augmentation.
(a) With data augmentation; (b) Without data augmentation.
Table 5.
The performance evaluation of the three proposed models against the models appeared in reference [20] and reference [21].
Fig 7.
Training progress of the proposed MResNet 3x3, 5x5, with and without data augmentation.
(a) With data augmentation; (b) Without data augmentation.
Fig 8.
Palm-leaves classification result using the proposed MResNet 3x3, 5x5, with and without data augmentation.
(a) With data augmentation; (b) Without data augmentation.
Fig 9.
The confusion matrix of the classification accuracy using the proposed MResNet 3x3, 5x5, with and without data augmentation.
(a) With data augmentation; (b) Without data augmentation.
Fig 10.
Training progress of the IncResNet.
Fig 11.
Palm-leaves classification result using the IncResNet.
Fig 12.
The confusion matrix of the classification accuracy using the IncResNet at 20 epochs.
Fig 13.
The performance evaluation of the three implemented models: ResNet, MResNet, and IncResNet.
(a) Validation accuracy; (b) Training accuracy; (c) Validation error; (d) Training error; (e) Overall recall; (f) Overall precision; (g) F1 score.
Fig 14.
Training and validation accuracy of the proposed models at epochs 10 and 30 compared with models in [20].
(a) Accuracy at epoch 10; (b) Accuracy at epoch 30.
Fig 15.
Validation accuracy, overall precision, overall recall, and F1 score of the proposed models compared with models in [21] at epoch 25.