Detection of Pepper Plant Leaf Disease Detection Using Tom and Jerry Algorithm With MSTNet

Leaf diseases have a detrimental effect on crop production quality in agriculture. In order to improve agricultural sector output, this has led to a greater emphasis on automating the detection of leaf diseases. In recent years, the process of categorising plant leaves using characteristics and machine learning has advanced. Usually, machine learning is used for supervised training of leaf classifiers using a set of data. Plant illnesses negatively impact both the amount and quality of agricultural goods by causing substantial growth and financial losses. Detecting plant diseases in big agricultural fields within a day has emerged as a critical topic of study. Three processes are used in this study to forecast pepper plant diseases (PPD): picture capture, feature selection, and image classification. The Kalman filter is used to remove noise from images before processing. Feature selection becomes important because it effectively solves the issue by eliminating redundant and unnecessary data, cutting down on computation time, improving learning accuracy, and improving comprehension of the model or data. For feature selection, Tom and Jerry optimisation (TJO) is used, and for final classification, the Modified Swin Transform method (MSTNet) is used. Using TJO, MSTNet's hyperparameters are adjusted to ascertain if a leaf is contaminated. On the Plant Village Dataset, experiments are carried out with different parameter measurements. The suggested MSTNet outperforms the accuracy rates of the current models with a 99.2% classification accuracy.

Перевод пока недоступен