Enhanced Disease Detection for Potato Crop Using CNN with Transfer Learning

Abstract

As the fourth most popular basic food in the world,potatoes are widely available. In addition, the worldwidemarket is causing the demand to rise daily. Diseases likeearly and late blight have a significant impact on the quantity and quality of potatoes. Determining which potato leaves are afflicted with a certain illness becomes more challenging when interpreting these diseasesmanually. Thankfully, it is possible to identify potato leafdiseases by examining the leaf conditions. This proposedstudy presents a technique that employs deep learning toidentify the two types of diseases and generates an accurate classifier using heavy designs for convolutionalneural networks, such as GoogleNet, Resnet15, VGG16,and Xception. We achieved 97% accuracy in the first 40 CNN epochs, demonstrating the practicality of the deep neural network approach.

Code and data availability

The paper's potato disease classification experiments are built directly on a public Kaggle PlantVillage image dataset (900 training / 300 validation images of Healthy, LateBlight, EarlyBlight potato leaves), which the authors explicitly reference with a public URL. No author code, models, or other paper-specific phenp