Strengthening Plant Disease Detection Using Ensemble Learning

Deep learning approaches have been applied to the agricultural sector to resolve several challenges, such as crop disease detection, yield prediction, irrigation management, etc., faced by farmers. Detecting and classifying plant diseases at an early stage is crucial for taking measures to prevent crop failure. In this paper, disease detection and classification of tomato plants is discussed. Experimentation involves transfer learning methods using deep learning models such as DenseNet201, InceptionV3, ResNet50, MobileNetV2, InceptionResNetV2, and Xception, along with a proposed ensemble learning model that integrates a voting ensemble of these models. The dataset contains 18,160 images divided into 10 classes. Among the individual models, DenseNet201 achieved the best accuracy at 94.54<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>whereas the proposed ensemble learning outperformed against the sate-of-the-art approaches.