Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Jurnal Keteknikan Pertanian Tropis dan Biosistem Advance Sustainable Science, Engineering and Technology (ASSET)
Nainggolan, Rut Juniar
Unknown Affiliation
Agriculture, Biological Sciences & Forestry Chemistry Computer Science & IT Control & Systems Engineering Electrical & Electronics Engineering Energy Industrial & Manufacturing Engineering Mechanical Engineering

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Prediction of Soil Nutrients from Different Soil Textures using Portable Spectrometer and Machine Learning Himawan, Harki; Nainggolan, Rut Juniar; Rakhmadi, Handono; Djoyowasito, Gunomo; Ubaidillah; Nopriani, Lenny Sri; Al Riza, Dimas Firmanda
Advance Sustainable Science Engineering and Technology Vol. 8 No. 1 (2026): November - January
Publisher : Science and Technology Research Centre Universitas PGRI Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26877/asset.v8i1.2166

Abstract

Soil nutrients, such as nitrogen, phosphorus, and potassium, are critical for plant growth and agricultural productivity. Conventional laboratory methods for measuring these nutrients are accurate but often time-consuming, costly, and environmentally taxing. This study explores the potential of portable visible-near infrared (Vis-NIR) spectrometer combined with machine learning algorithms as a rapid, cost-effective, and eco-friendly alternative for soil nutrient analysis. Soil samples of clay, clay loam, and sandy clay were collected and analyzed using artificial neural network (ANN) approach to predict soil nutrients. A total of 81 reflectance spectra data from each soil type were acquired using an AS7265x sensor and processed to develop a predictive model for nutrient content. ANN models demonstrated high accuracy, with R² values exceeding 0.8 in each type of soil texture. This study emphasizes the potential of portable Vis-NIR spectrometer and machine learning integration to revolutionize soil nutrient analysis, offering significant improvements in agricultural efficiency and sustainability.
Optimized ResNet-18 Model for Ripeness Classification of Javanese Long Pepper (Piper retrofractum) Using Reflectance and Fluorescence Imaging Sandra; Damayanti, Retno; Sa'diyah, Mitha; Nainggolan, Rut Juniar
Jurnal Keteknikan Pertanian Tropis dan Biosistem Vol. 14 No. 1 (2026): April 2026
Publisher : Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21776/

Abstract

Javanese long pepper (Piper retrofractum Vahl.) is a high-valuable horticultural commodity crop whose quality and bioactive composition depend strongly on fruit ripeness. Conventional visual assessment is often subjective and inconsistent, emphasizing the need for an objective and non-destructive alternative. This study aimed to develop an automated classification system for determining ripeness stages (green, orange, and red) using reflectance (visible light) and fluorescence (ultraviolet) imaging analyzed through a fine-tuned ResNet-18 deep learning model. A total of 300 images per modality were captured under controlled studio conditions and processed through standardized background removal, size normalization, and data augmentation. Color feature analysis across RGB, CIELAB, and HSV color spaces revealed that reflectance imaging effectively captured pigment transitions associated with chlorophyll degradation and carotenoid accumulation, with the a* channel serving as the most discriminative single-channel indicator. Fluorescence imaging provided complementary physiological information through chlorophyll emission dynamics, exhibiting greater inter-class overlap particularly between green and orange stages. The ResNet-18 model was evaluated using three optimizers (Adam, SGDM, and RMSprop) based on accuracy, precision, recall, Macro F1-score, and loss. For reflectance imaging, SGDM and RMSprop both achieved perfect test classification (accuracy = 100%, Macro F1 = 1.000), while fluorescence-based models achieved up to 97.78% accuracy. Training curve analysis confirmed stable convergence without overfitting across all combinations. Confusion matrix analysis showed that misclassifications were confined to adjacent ripeness stage boundaries, and Grad-CAM visualization confirmed physiologically consistent spatial attention patterns in the best-performing models. The proposed imaging and modeling pipeline offered a reproducible framework for postharvest quality evaluation and standardization of Piper retrofractum and similar horticultural commodities.
Co-Authors Dimas Firmanda Al Riza Gunomo Djoyowasito Himawan, Harki Rakhmadi, Handono Retno Damayanti Sa'diyah, Mitha Sandra Ubaidillah