Garuda - Garba Rujukan Digital

Comparative Study of Mobilenet and Resnet for Watermelon Leaf Disease Classification Using Deep Learning Ahmad, Abdullah; Wanto, Anjar; Windarto, Agus Perdana; Poningsih, Poningsih
TIN: Terapan Informatika Nusantara Vol 6 No 1 (2025): June 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/tin.v6i1.7543

Watermelon leaf diseases, caused by various factors such as fungi, viruses, and bacteria, can have a significant impact on agricultural yields. To increase the amount and quality of watermelon produced, early diagnosis of this disease is essential. This study aims to compare the performance of two Convolutional Neural Networks (CNN) architectures included in Deep Learning, namely MobileNet and ResNet, in classifying watermelon leaf diseases using a dataset taken from Kaggle. This dataset consists of 1000 watermelon leaf images with three conditions, namely Downy Mildew (380 images), Healthy (205 images), and Mosaic Virus (415 images). ). 95% accuracy, 96% precision, 94% recall, and 95% f1-score are the results of the MobileNet model. In contrast, the ResNet model performs better, with 97% accuracy, 96% precision, 97% recall, and 97% f1-score. The study's findings show that ResNet outperforms MobileNet in the classification of watermelon leaf illnesses, despite both models' excellent and effective performance for automatic plant disease detection applications.

OPTIMIZING SHUFFLENET WITH GRIDSEARCHCV FOR GEOSPATIAL DISASTER MAPPING IN INDONESIA Ahmad, Abdullah; Hartama, Dedy; Solikhun, Solikhun; Poningsih, Poningsih
JITK (Jurnal Ilmu Pengetahuan dan Teknologi Komputer) Vol. 11 No. 2 (2025): JITK Issue November 2025
Publisher : LPPM Nusa Mandiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33480/jitk.v11i2.6747

Accurate classification of natural disasters is crucial for timely response and effective mitigation. However, conventional approaches often suffer from inefficiency and limited reliability, highlighting the need for automated deep learning solutions. This study proposes an optimized Convolutional Neural Network (CNN) based on the lightweight ShuffleNet architecture, enhanced through GridSearchCV for systematic hyperparameter tuning. Using a geospatial dataset of 3,667 images representing earthquake, flood, and wind-related disasters in Indonesia, the optimized ShuffleNet model achieved a peak accuracy of 99.97%, outperforming baseline CNNs such as MobileNet, GoogleNet, ResNet, DenseNet, and standard ShuffleNet. While these results demonstrate the potential of combining lightweight architectures with automated optimization, the exceptionally high performance also indicates possible risks of overfitting and dataset bias due to limited variability. Therefore, future research should validate this approach using larger, multi-source datasets to ensure robustness and real-world applicability

OPTIMIZATION OF SVM ALGORITHM FOR OBESITY CLASSIFICATION WITH SMOTE TECHNIQUE AND HYPERPARAMETER TUNING Nur, Khairun Nisa Arifin; Wanto, Anjar; Poningsih, Poningsih
JITK (Jurnal Ilmu Pengetahuan dan Teknologi Komputer) Vol. 11 No. 2 (2025): JITK Issue November 2025
Publisher : LPPM Nusa Mandiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33480/jitk.v11i2.6878

Excessive fat accumulation that impairs personal health and raises the risk of chronic diseases is the hallmark of obesity, a global health issue. Decision Tree (DT) has been widely used for obesity classification, but it tends to suffer from overfitting and poor performance on imbalanced datasets. To overcome these limitations, this study proposes an optimization of the Support Vector Machine (SVM) algorithm using Synthetic Minority Over-sampling Technique (SMOTE) and Hyperparameter Tuning. SMOTE was applied to balance the class distribution, whereas Grid Search was utilized to determine the optimal combination of hyperparameters (C, gamma, and kernel). The dataset employed in this research comprises multiple features related to individual health and lifestyle, with obesity level as the target class. The experimental results demonstrate that the optimized SVM model demonstrated strong classification performance, attaining 97% in accuracy, precision, recall, and F1-score. This high performance is significant because it enables more accurate early detection of obesity risk, which can support timely medical intervention and personalized treatment planning, ultimately contributing to better public health outcomesThese findings suggest that incorporating SMOTE and Hyperparameter Tuning substantially improves SVM performance, establishing it as a robust approach for obesity classification and early risk detection.

OPTIMIZING SHUFFLENET WITH GRIDSEARCHCV FOR GEOSPATIAL DISASTER MAPPING IN INDONESIA Ahmad, Abdullah; Hartama, Dedy; Solikhun, Solikhun; Poningsih, Poningsih
JITK (Jurnal Ilmu Pengetahuan dan Teknologi Komputer) Vol. 11 No. 2 (2025): JITK Issue November 2025
Publisher : LPPM Nusa Mandiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33480/jitk.v11i2.6747

Accurate classification of natural disasters is crucial for timely response and effective mitigation. However, conventional approaches often suffer from inefficiency and limited reliability, highlighting the need for automated deep learning solutions. This study proposes an optimized Convolutional Neural Network (CNN) based on the lightweight ShuffleNet architecture, enhanced through GridSearchCV for systematic hyperparameter tuning. Using a geospatial dataset of 3,667 images representing earthquake, flood, and wind-related disasters in Indonesia, the optimized ShuffleNet model achieved a peak accuracy of 99.97%, outperforming baseline CNNs such as MobileNet, GoogleNet, ResNet, DenseNet, and standard ShuffleNet. While these results demonstrate the potential of combining lightweight architectures with automated optimization, the exceptionally high performance also indicates possible risks of overfitting and dataset bias due to limited variability. Therefore, future research should validate this approach using larger, multi-source datasets to ensure robustness and real-world applicability

OPTIMIZATION OF SVM ALGORITHM FOR OBESITY CLASSIFICATION WITH SMOTE TECHNIQUE AND HYPERPARAMETER TUNING Nur, Khairun Nisa Arifin; Wanto, Anjar; Poningsih, Poningsih
JITK (Jurnal Ilmu Pengetahuan dan Teknologi Komputer) Vol. 11 No. 2 (2025): JITK Issue November 2025
Publisher : LPPM Nusa Mandiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33480/jitk.v11i2.6878

Excessive fat accumulation that impairs personal health and raises the risk of chronic diseases is the hallmark of obesity, a global health issue. Decision Tree (DT) has been widely used for obesity classification, but it tends to suffer from overfitting and poor performance on imbalanced datasets. To overcome these limitations, this study proposes an optimization of the Support Vector Machine (SVM) algorithm using Synthetic Minority Over-sampling Technique (SMOTE) and Hyperparameter Tuning. SMOTE was applied to balance the class distribution, whereas Grid Search was utilized to determine the optimal combination of hyperparameters (C, gamma, and kernel). The dataset employed in this research comprises multiple features related to individual health and lifestyle, with obesity level as the target class. The experimental results demonstrate that the optimized SVM model demonstrated strong classification performance, attaining 97% in accuracy, precision, recall, and F1-score. This high performance is significant because it enables more accurate early detection of obesity risk, which can support timely medical intervention and personalized treatment planning, ultimately contributing to better public health outcomesThese findings suggest that incorporating SMOTE and Hyperparameter Tuning substantially improves SVM performance, establishing it as a robust approach for obesity classification and early risk detection.

Title

Found 25 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 25 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 25 Documents
Search