Detection of Potato Leaves Using Convolutional Neural Networks (CNNs)

Agricultural productivity plays a vital role in India’s economy, contributing approximately 17–18% to the nation’s GDP and serving as a primary livelihood for a significant portion of the population. With India possessing the largest net cropped area globally, effective pest and disease management is crucial to ensuring stable agricultural yields. This study utilizes advanced technologies, including machine learning, computer vision, and deep learning, to detect and identify common diseases such as early blight and late blight in potato plant leaves. Using a Kaggle-sourced dataset comprising roughly 2,000 images, rigorous data augmentation techniques were applied to enhance the robustness of the model. A customized convolutional neural network (CNN) was deployed, achieving an impressive accuracy of 97.12% after just 32 epochs. This performance surpasses many existing Kaggle models, which typically reach around 95% accuracy.