<![CDATA[This study aims to develop a model for estimating nutrient content (ppm) in spinach plants using the YOLOOD architecture. Nutrient estimation is performed based on leaf image analysis as a non-destructive approach to detect nutrient deficiencies at an early stage. The method involves collecting spinach leaf images with six nutrient variation levels (100, 300, 500, 700, 900, and 1200 ppm), with 300 images per class, followed by annotation, augmentation, preprocessing, and dataset splitting into training and validation sets with a 70:30 ratio. The model is trained for 50 epochs with a batch size of 4, an input image size of 416×416 pixels, the Adam optimizer, and a learning rate of 0.0001 using default YOLOOD parameters. The model is designed to recognize visual differences in spinach leaves across nutrient levels and estimate nutrient concentration values. Performance evaluation is conducted using Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE), and the coefficient of determination (R²). The results indicate that the model achieves good object detection performance with an mAP@50 of 0.93; however, in the nutrient estimation stage, it obtains a MAPE of 16%, a normalized RMSE of 0.1571, and an R² of 0.7903. Therefore, the YOLOOD model is considered effective in detecting visual characteristics of spinach leaves and reasonably capable of estimating nutrient content, although further improvements are needed to enhance prediction accuracy.]]>
Copyrights © 2026
