ABSTRACT
The purpose of this research is to use machine learning approaches to detect early and late blight in tomato as well as potato infections so as to develop treatments for them. We selected Kaggle's Plant Village dataset for this study since it encompasses 3500 images of the infected plants, including 1700 images for tomatoes and 1800 images for potatoes. The photos for the potatoes are situated in one half of the dataset, while the images for the tomatoes are in the other half. We used various methods in our research: dataset normalization and split, data augmentation and image resizing, feature extraction and classification, and sickness forecasting. The machine learning approaches for the sickness prediction are trio algorithms: K-Nearest Neighbours , Random Forest, and Artificial Neural Networks . According to the test outcomes, tomatoes and potatoes are excellent sickness indicators. The accuracy for the K-NN model amounts to 96.4%, the one for the ANN exceeds this result and equals 98.5%, while for the Random Forest, it is 97.8%. This information implies that machine learning systems can perfectly classify and identify plant sicknesses. The immediate implications of this study have to do with the potential agricultural methods of the future. If farmers can tell if a plant is sick early on, they can prevent disease from spreading, thus, avoiding substantial crop losses. The outcomes demonstrate how the ML technology can assist in disease suppression in potatoes and tomatoes as well as in a variety of agricultural possibilities.
