Machine Learning in Agriculture: Enhancing Productivity, Sustainability, and Resilience

Abstract

Machine learning (ML) is quickly revolutionizing agriculture by allowing the use of data to help make evidence-based decisions that lead to more production, less waste, and more sustainability. The paper discusses the essence of ML techniques in the agriculture industry, iconic examples of their usage (crop yield prediction, pest and disease detection, precision irrigation, and supply-chain optimization), presents a generalized approach to the implementation of ML in an agribusiness, cites limitations to its application (data quality, model generalizability, and socio-economic barriers), and suggests future research directions. We believe that integrating ML with low-cost sensing, participatory data collection, and domain-intelligent models can enable significant productivity benefits to both smallholder and commercial farms and reduce the negative environmental impact.