In this blog, we show how neural networks can be used for plant disease recognition in the context of image classification. We used the publicly available Plant Village dataset which has cotton-related diseases. We compared five different backbones including VGG16, ResNet50, InceptionV3, InceptionResNet and DenseNet169. We found that ResNet50 achieves the best result on the test set. For evaluation, we used metrics: accuracy, precision, recall, F1 score, and class-wise confusion metric. Our model achieves the best of results using ResNet50 with an accuracy of 0.982, the precision of 0.94, recall of 0.94, and F1 score of 0.94.
DATA SET:
A dataset contains 1951 images of disease and healthy cotton plants leave under controlled conditions. The images are further divided into 3 parts: 1)Test
2)Train
3)Validity
All three categories are further divided into four parts:
1)Disease cotton plant
2)Disease cotton leaves
3)Fresh cotton plant
4)Fresh cotton leaves
There are mainly 5 types of cotton disease:
1) Alternaria leaf spot alternaria macrospora.
2) Asochyta blight Asochyta gossypii.
3) Cercospora leaf spot Cercospora gossypina.
4) Fusarium wilt Fusarium oxysporum.
5) Target spot Corynespora cassiicola.
In the proposed work there are 1614 images in the training data set, 319 images in a validation data set, and 18 images in the testing data set. Out of 1614 training data images, 807 images belonging to the healthy category, and 807 images belonging to each cotton disease category described above.
MACHINE LEARNING:
Machine learning is a method of data analysis that automates analytical model building. It is a branch of artificial intelligence based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention.
