ENHANCING POTATO CROP HEALTH MONITORING USING TWO STAGE CNN BASED DISEASE CLASSIFICATION MODELS

Potato is one of the most extensively cultivated crops in India as well as worldwide andserves as a staple food in many regions. Due to its agricultural and economic importance, effectivedisease management is essential to ensure healthy crop yields. Traditional methods of disease detectionrely on manual visual inspection by farmers or agricultural experts, which often lack precision and areprone to misdiagnosis, leading to substantial crop losses. This study proposes a deep learning-basedapproach for automated disease detection in potato plants using a Convolutional Neural Network(CNN). The novelty of this work lies in developing a two-stage CNN framework that first classifiesthe potato leaf images as either healthy or unhealthy. Next, the unhealthy leaf images are furtherclassified as either early blight or late blight diseases. The proposed CNN models are optimized usingthe Adam optimizer with a learning rate of 0.0001. Extensive experimentation on a dataset of 1,500images demonstrates high classification accuracies of 98.3% and 99% for the two stages, respectively.The results demonstrate that our proposed CNN models are highly effective for automated diseasedetection and improve decision-making in potato crops.