Ensembling of Transfer Learning for Enhanced Precision Agriculture in Plant Disease Classification

Precision agriculture utilizes advanced technologies and data-driven insights to improve agricultural yields and resource efficiency. This paper proposes ensemble models that incorporate various pretrained deep learning architectures to improve the precision of automated disease detection, promoting efficient and sustainable farming practices. Two ensembles, one with DenseNet121, VGG19, and ResNet50, and the other with MobileNet, InceptionV3, and Xception, have been developed and evaluated on a dataset of healthy, powdery, and rusty plant pictures. The second ensemble demonstrated 97% accuracy, indicating its efficacy in automated plant leaf disease classification. This study promotes precision agriculture by establishing a reliable framework for precise disease identification, enabling timely interventions, and increasing crop monitoring for sustainable farming practices.