Identification of Macro-Nutrient Deficiency in Onion Leaves (Allium cepa L.) Using Convolutional Neural Network (CNN)
Kristin Clarisse H. Mateo, Laurence Kobe B. Navarro, Cyrel O. Manlises
Abstract
In the Philippines, the onion is among the major crops produced with an estimated production of 229.50 thousand metric tons last 2020. Producing these onions relies on manpower, particularly from local farmers. However, the nutrient deficiency that the onions may have can affect the product quality and crop yield. The possible nutrient deficiency of onions may be identified based on the features of the leaves. Therefore, based on the visual symptoms, a modern approach to identifying the primary macro-nutrient deficiency in onion leaves was developed using a Convolutional Neural Network with the VGG-16 architecture. The system obtained an accuracy of 60%, 76%, and 100% in identifying the macro-nutrient deficiency of onion leaves in Nitrogen, Phosphorus, and Potassium, respectively. It obtained an accuracy of 96% in determining no nutrient deficiency in onion leaves. The overall system accuracy obtained was 83%.