Fertilizer Optimization through Machine Learning-driven Models: An Empirical Investigation on Smart Farming of Amaranth

Abstract

Amaranth is a highly nutritious leafy vegetable cum pseudo-cereal crop known for its adaptability to various climatic conditions, making it a promising crop for addressing the food security and nutritional needs of a growing population. To enhance its quality and boost yields, farmers mainly depend on synthetic fertilizers. However, the excessive use of inorganic fertilizers to maximize crop yields poses significant ecological risks. This study aimed to investigate the impact of excessive inorganic fertilizer on the growth, yield, and physiological attributes of Amaranthus with the aid of advanced machine learning paradigm. An experimental pot trial was conducted using different NPK fertilizer dosage regimes, and agronomic parameters such as moisture level, crop yield, plant height, leaf length, leaf width etc. were measured and analyzed using statistical methods. The results demonstrated that the application of excessive inorganic fertilizer initially promoted plant growth but surpassed optimal levels resulting in negative effects, including stunted growth and reduced vigor. By identifying the Amaranthus's productivity and adaptability in different chemically treated soil conditions and automatically phenotyping its traits using image-based machine learning models, this study aims to determine the overuse of synthetic fertilizers. A comparative evaluation of different learning algorithms was carried out and the experimental result proves that SVM classifier could be a more appropriate learning algorithm for the proposed system with 80% accuracy. These findings highlight the importance of adopting sustainable fertilizer practices for the cultivation of Amaranthus and emphasize the need for ecological balance in crop production systems.