GREENLINK: ADVANCING PLANT GROWTH WITH AI

Abstract

The research is divided into two parts - The first part to the research involves Integrat ing K-Fold Cross-Validation with Convolutional Neural Networks (CNN) for Plant Species and Pathogen Detection, that focuses on precisely identifying and predicting various plant species using Artificial Intelligence (AI) techniques like CNN and K-fold Cross-Validation and also accurately diagnosing the disease that the plant under consideration is affected by. In the research, we utilized a rich dataset from the PlantVillage repository, and our models were trained on over 54,306 images that also cover 14 major crop species. The model identifies the plant and pathogens and then focuses on the accuracy of identifying the right kind of species and pathogens. The growth prediction model predicts the best conditions for the plant to grow. In the work, the results were successfully tested and witnessed 81 % accuracy in the Plant and Pathogen detection model and the growth prediction model’s low mean squared error i.e. 21 % supports accurate trend forecasting for optimizing plant care. The second part to the research contributes towards Opti mising Plant Health with Q-Learning, that introduces a novel approach to plant care, leveraging deep reinforcement learning (DRL) algorithms, such as Q-learning, to simu late diverse plant growth scenarios. The research aims to develop a system that provides a tailored approach to determine the best-case scenario for plant species’ maximum or optimum growth or development. The PlantVillage dataset used for the research is well labeled and considered as it fulfills the set environment for the agent to produce rewards on. The research contributes significantly to environmental sustainability and ecological awareness, fostering a deeper connection between humans and the natural environment by providing AI-powered cultivation strategies.