Detection of Nutrient Deficiencies in Coffea Arabica Leaves Using the YOLO Object Detection Algorithm

Abstract

Purpose – Nutrient deficiencies in coffee plants can severely impact plant health and yield, making timely detection crucial for farmers. Traditional methods, such as visual examination and soil testing, are inefficient and lack scalability. This study aims to develop a model using YOLOv7 to detect multiple nutrient deficiencies in Coffea arabica leaves, offering a practical and real-time solution for coffee farmers to monitor plant health and optimize yield.

Method – A dataset of 462 images of coffee leaves showing nutrient deficiencies (Potassium, Magnesium, Manganese, Sulfur, Copper, Molybdenum, Iron, and Zinc) was compiled, with labels verified by experts. Two YOLOv7 models were trained for 100 epochs: a fixed resolution model and a multi-resolution model. A prototype system was also developed to detect these deficiencies in real time for end-users.

Results – The multi-resolution YOLOv7 model achieved 75% accuracy, surpassing expectations given the limited dataset size. The multi-resolution model outperformed the fixed-resolution model, demonstrating the efficacy of this approach in detecting multiple nutrient deficiencies simultaneously.

Conclusion – The study confirmed the potential of YOLOv7 in real-time detection of multiple nutrient deficiencies in coffee leaves, providing an efficient, scalable solution for farmers. This technology could enable timely interventions to improve crop health and yield.

Recommendation – Further expansion of the dataset and real-world testing are recommended to validate and improve the model’s accuracy. Exploration of additional learning models and techniques could further enhance detection performance.

Practical Implications – This research highlights the potential of integrating deep learning models, like YOLOv7, in agricultural monitoring systems. The developed prototype provides farmers with a tool to optimize plant health management, showcasing the broader impact of AI technology in improving efficiency and productivity in agriculture.