AI-Powered Potato Disease Classification using ResNet50 for Detecting Late Blight in Crop Leaves

The agricultural world produces potato as its most widely grown food crop because it maintains important roles in both food security and economic stability. Agricultural production of potato faces severe risk due to late blight infections because these diseases can lead to substantial yield reductions unless professionals identify them promptly for early intervention. Human-caused disease detection methods are slow and error-prone since people restrict their accuracy level through these processes. The development of artificial intelligence systems enables deep learning through convolutional neural networks (CNNs) to automate plant disease detection. This study investigates the effectiveness of implementing a ResNet50 model that underwent fine-tuning to distinguish potato plant leaves between healthy and late blight infected ones. For the experimental analysis the research team employed 430 high-resolution images that were divided into 302 training images and 128 validation images after performing preprocessing steps combined with input requirements resizing for ResNet50. Pre-trained weights from ImageNet served as a basis under transfer learning methods to enhance performance. Training occurred during 20 epochs through the implementation of Adam optimizer with a batch size set to 64. The model reached a validation accuracy of 88% through its analysis of the provided information while maintaining high precision and recall for identifying healthy samples. Deep learning-based models have proven their ability to precisely identify potato leaf diseases through these findings thus providing a dependable tool for early diagnosis within smart agricultural systems.