Advancing Agricultural Sustainability: Smart Monitoring of Oil Palm Seedlings Through Innovative Image Processing Techniques

Abstract

Good oil palm seedlings planted in Malaysia can enhance the nursery area, the economy, and rural employment. Nitrogen (N), potassium (K), \and magnesium (Mg) deficiencies in oil palm seedlings could have an adverse effect on growth and seedling quality. In contrast, excess fertiliser in oil palm seedlings could reduce macronutrients and soil organic matter levels. Moreover, various diseases appear on the oil palm seedling leaves resulting from nutrient deficiencies. Oil palm seedlings with nutrient deficiencies were more susceptible to pathogens and diseases than healthy oil palm seedlings. Thus, image processing made it possible to quickly and accurately control oil palm seedlings' growth and avoid diseases. In this research study, image processing was proposed to classify nutrient deficiencies in oil palm seedling leaves subjected to the three different fertiliser rates. The dataset for this research study was collected from a nursery and consisted of 868 images classified into four classes (Healthy, Nitrogen, Potassium and Magnesium). According to the experiment's findings, the Xception model achieved the highest percentage of classification accuracy, 98.60%, within a short time. It can be concluded that the proposed implementation of image processing for the classification of nutrient deficiencies in oil palm seedling leaves was effective. However, more datasets could be added in the future to achieve a better balance and enhance classification performance.