Recurrent Neural Network-Based Classification of Potato Leaves using RGB Images

In this study, a highly accurate classification system for potato leaves using Recurrent Neural Networks (RNNs) on the Plant Village dataset has been presented. The dataset comprises a diverse collection of potato leaf images, with annotations for healthy leaves and those affected by various diseases and stressors. Leveraging the temporal dependencies inherent in RNNs, it is aimed to effectively capture intricate patterns and features in sequential data, particularly crucial for time-series analysis of plant growth stages and disease progression. The proposed RNN architecture, incorporating long short-term memory (LSTM) units to address the vanishing gradient problem, demonstrated exceptional performance in accurately classifying potato leaf health, enabling early disease detection and timely interventions for improved crop management. Our study demonstrates the robustness and high accuracy of RNNs in classifying potato leaves, with performance metrics exceeding 92%. The integration of RNN-based classification systems into precision agriculture holds tremendous promise, providing farmers with valuable insights for timely interventions, optimizing crop health, and ensuring sustainable food production. From the experimental outcomes, it has been observed that the “RNN Model“ records the highest accuracy of 0.927, which is a higher value compared to the “CNN (Convolution neural network) Model“ (0.902) and the “Feedforward neural network“ (0.867). The success of proposed DL model in agricultural applications illustrates the transformative potential of AI technologies in addressing global food security challenges and revolutionizing the future of agriculture.