STRAWBERRY PLANT DISEASE DETECTION USING U-NET SEGMENTATION AND XGBOOST CLASSIFICATION
ID：8
View protection：Participant Only
Updated time：2025-11-04 14:05:01 Views：184
Oral (Online)
Abstract
Despite being one of the horticultural crops with the highest economic value, strawberries are extremely vulnerable to bacterial and fungal infections. To reduce yield losses and guarantee sustainable cultivation, early and precise disease detection is crucial. This is also the first research where an integrated technique that uses machine learning and deep learning to classify diseases of strawberry plants is developed. A U-Net segmentation model is first used to determine the locations of diseased strawberry plants, based on the images of their leaves which the model has been trained on. During training, K-fold cross-validation is utilized to ensure the model is optimally trained such that it can generalize on new samples. After some extra steps, special features that help in distinguishing between different classes are taken from the segmented results to perform the classification.An XGBoost classifier is then used to categorize the leaves into classes that are either healthy or diseased. K-Means clustering is used as a preprocessing step for better lesion feature extraction. Based on experimental results, the suggested framework provides 92% classification accuracy and roughly 90% segmentation accuracy (Dice coefficient). Due to its effectiveness and dependability in computerized disease detection in strawberry plantations, this combination approach holds great potential for application in precision agriculture.
Keywords
Deep Learning, Machine Learning, Computer Vision, XGBoost, K-Means Clustering, Image Segmentation, Strawberry Leaf Disease, Convolutional Neural Network (CNN), U-Net, Precision Agriculture.
Post comments