Article Per Year (5 Year)
p-Index From 2021 - 2026
1.556
P-Index
This Author published in this journals
All Journal InfoTekJar : Jurnal Nasional Informatika dan Teknologi Jaringan Sinkron : Jurnal dan Penelitian Teknik Informatika Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) JURNAL MEDIA INFORMATIKA BUDIDARMA Journal of Information Systems and Informatics Indonesian Journal of Electrical Engineering and Computer Science Innovation in Research of Informatics (INNOVATICS)
Herimanto Herimanto
Institut Teknologi Del
Computer Science & IT Control & Systems Engineering Electrical & Electronics Engineering Engineering

Published : 8 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 8 Documents
Search

 1 
Improving Performance of The Genetic Algorithm on NP-Complete Problem Herimanto Herimanto; Muhammad Zarlis; Syahril Efendi
InfoTekJar : Jurnal Nasional Informatika dan Teknologi Jaringan Vol 6, No 1 (2021): InfoTekJar September
Publisher : Universitas Islam Sumatera Utara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30743/infotekjar.v6i1.3909

Abstract

Non-Deterministic Polynomial Complete Problem is the most challenging problem and also engaging in algorithm strategy. One representation of this problem is the sudoku numbers game. To fill an empty sudoku puzzle, a specific formula does not apply, but filling in sudoku is a matter of decision. So it takes a special algorithm and strategy to solve it. As such, the case of the sudoku numbers game has been widely praised as the topic of finding the best results. One of the methods used is a genetic algorithm. However, due to many processes and data used in the implementation of genetic algorithms, the results obtained are often not optimal. This research will introduce a special strategy in implementing genetic algorithms in NP-Complete problems, namely by optimizing the genetic algorithm in the process of population formation. From the test results, it is found that the application of the genetic algorithm with optimization results in smaller time data and test data compared to the algorithm without optimization.
Perbandingan Matriks Loss Pada Model Deep Learning Resnet50 dan Xception dalam Deteksi Objek Herimanto Herimanto
JURNAL MEDIA INFORMATIKA BUDIDARMA Vol 7, No 4 (2023): Oktober 2023
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/mib.v7i4.6849

Abstract

The implementation of deep learning has expanded into various fields, not confined solely to the field of education, particularly in computer science. It has also integrated technology into various other domains, including geospatial, remote sensing, and even the medical field. This development has made a significant contribution to reshaping the way humans understand and tackle challenges across different sectors. In this context, deep learning is employed for object detection and classification. Despite the considerable progress facilitated by the application of deep learning, object detection remains a challenge that is not entirely resolved. Constraints such as variations in lighting conditions, angles of view, and object diversity make achieving high-accuracy object detection a difficult task. Therefore, further research is required to comprehend and compare the performance of various deep learning models in addressing this issue. This research focuses on the comparison of two deep learning models, namely ResNet50 and Xception, in terms of loss metrics when detecting an object, in this case, a chair. The models are provided with input images of chairs and predict whether the chairs are empty or occupied. The results obtained from this research indicate that the ResNet50 model has a lower total loss value of 0.19422098, while the Xception model has a total loss value of 1.1822930. The lower the loss value, the better the model's performance. Based on the comparison results, the author has developed a web application simulator using Flask, utilizing the model with the lowest loss, which is the ResNet50 model.
A Comparative Analysis of Content-Based Filtering and TF-IDF Approaches for Enhancing Sports Recommendation Systems Herimanto, Herimanto; Samosir, Kevin; Ginting, Fastoria
Innovation in Research of Informatics (Innovatics) Vol 6, No 2 (2024): September 2024
Publisher : Department of Informatics, Siliwangi University, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37058/innovatics.v6i2.12404

Abstract

Sport is one of the important factors for someone to maintain or improve their health. There are many purposes for a person to exercise. However, many people still find it difficult to determine the relevant type of exercise according to their preferences. Recommendation systems are now an important element in human life to provide relevant recommendations to users. The purpose of this research is to develop a sports recommendation system that can provide accurate types of sports recommendations to users and describe how the recommendation system can work in providing recommendations when cold-start problems and non-cold-start problems occur. The method used in this research is content-based filtering by applying Term Frequency - Inverse Document Frequency (TF-IDF) vectorization matrix and cosine similarity algorithm. When a new user logs in, the system first checks the user's preferences to determine whether a cold-start problem or non-cold-start problem occurs. When a cold-start problem occurs, TF-IDF will be used in providing recommendations to the user. Conversely, when a non-cold-start problem occurs, cosine similarity will be used. The results show that by using TF-IDF and cosine similarity, the system successfully provides relevant sports recommendations to users in both cold-start problem and non cold-start problem situations with an accuracy rate of 86.90%. The novelty of this research lies in the understanding of sports provided to users through sports-related journals. Through these journals, it can increase user satisfaction, trust, compliance, and educate users in running sports
Enhancing Real-time Herbal Plant Detection in Agricultural Environments with YOLOv8 Siahaan, Ranty Deviana; Pardede, Herimanto; Simbolon, Iustisia Natalia; Simbolon, Ivanston; Gultom, Dian Jorgy
Journal of Information System and Informatics Vol 6 No 4 (2024): December
Publisher : Universitas Bina Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51519/journalisi.v6i4.889

Abstract

The detection of herbal plants plays a crucial role in the utilization of traditional medicine, particularly in the Toba region of Indonesia. This study aims to develop an Android application capable of real-time detection of herbal plants using the YOLOv8 algorithm. The five types of herbal plants targeted in this study are tempuyung, rimbang, papaya leaves, turmeric leaves, and aloe vera. The research methodology includes the collection of a dataset of herbal plant images, which were then labeled using the Roboflow platform. The YOLOv8 model was trained with this dataset to detect herbal plant objects. After training, the model was exported to TensorFlow Lite and integrated into an Android application. Testing was conducted to evaluate the accuracy and real-time detection performance of the application. The results show that the YOLOv8 model achieved a mean Average Precision (mAP) of 92.4%, with optimal real-time detection capabilities on Android devices. The developed application can quickly and accurately detect and identify herbal plants, providing a practical solution for users to recognize herbal plants. This study indicates that the YOLOv8 algorithm is effective for herbal plant recognition applications in a mobile context, opening up opportunities for further development in the integration of AI technology into everyday applications.
Performance Analysis of MobileNetV3-based Convolutional Neural Network for Facial Skin Disorder Classification Herimanto; Arie Satia Dharma; Junita Amalia; David Largo; Christin Adelia Pratiwi Sihite
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 8 No 6 (2024): December 2024
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v8i6.5982

Abstract

Accurately identifying facial skin types is essential for recommending the right skincare treatments and products. Misidentifying skin types can lead to negative consequences, such as irritation or worsening of skin conditions. This study investigated methods for classifying facial skin types into five categories: oily, acne-prone, dry, normal, and combination. A dataset of 1725 augmented facial images was used. Data augmentation techniques likely increased the dataset's diversity, which helps improve the model's generalization ability. The data underwent preprocessing, including rescaling, before being applied to two deep learning models, CNN and MobileNetV3. The models were evaluated based on accuracy and execution time to determine the most effective approach for classifying facial skin types. The CNN model achieved an accuracy of 64%, demonstrating its potential for image classification tasks. However, the MobileNetV3 model significantly outperformed CNN with an accuracy of 84%. This superior performance is attributed to MobileNetV3's advanced architecture, which is optimized for efficient feature extraction, and particularly relevant for capturing the subtle variations in facial skin types. Therefore, MobileNetV3 emerged as the more effective method for classifying facial skin types with higher accuracy.
Development of mobile-based Batak script recognition application using YOLOv8 algorithm Simbolon, Iustisia Natalia; Herimanto, Herimanto; Siahaan, Ranty Deviana; Lumbantobing, Samuel Adika; Br Sitepu, Grace Natalia
Indonesian Journal of Electrical Engineering and Computer Science Vol 39, No 2: August 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijeecs.v39.i2.pp1013-1026

Abstract

The Batak people are one of the ethnic groups that pass down many values and traditions to each generation, including the written tradition known as the Batak script. The Batak Toba people, in particular, have the Batak Toba script as part of their local wisdom that needs to be preserved and maintained. However, the use of the Batak script has significantly declined in the current era. To prevent the loss of this heritage, preservation through technology is necessary. This research utilizes a deep learning approach using the YOLOv8 algorithm to detect images of script objects, provide the coordinates of the script locations, and perform object recognition based on the dataset. The final result of this research is an Android-based application that can detect the Batak Toba script in real time and upload images. The research process involves experiments on several hyperparameters, such as epochs with a value of 200, confidence threshold, and IoU with a value of 0.5. The model evaluation shows excellent results, with a precision of 0.945, recall of 0.902, mAP@0.5 of 0.954, and a high confidence score from the application's detection.
Performance Trade-off of Anchor-Based and Anchor-Free Approaches of Faster R-CNN for Face Detection Dharma, Arie Satia; Herimanto, Herimanto; Simbolon, Niar Fujita; Nababan, Anton Roycar
Sinkron : jurnal dan penelitian teknik informatika Vol. 10 No. 1 (2026): Article Research January 2026
Publisher : Politeknik Ganesha Medan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33395/sinkron.v10i1.15804

Abstract

The face is a unique biometric feature that plays a crucial role in individual identification as it holds essential information for identity recognition. Face detection technology has been experiencing significant advancements in the field of computer vision. However, face detection technology continues to face challenges in balancing high detection accuracy with computational efficiency. While deep learning has advanced this field, there remains a lack of comparative studies that compare the performance trade-offs between anchor-based and anchor-free region proposal mechanisms within a Faster R-CNN framework. This research objective is comparing the performance of face detection using two approaches: anchor-based and anchor-free. The anchor-based approach use anchor boxes to predict bounding boxes, while the anchor-free approach predicts bounding boxes directly from pixel positions oriented around a point. The anchor-based approach is implemented use base line region proposed network method, whereas the anchor-free approach use a centerpoint method. The study utilizes a custom dataset comprising 1,000 formal images of students from Del Institute of Technology, split into 900 training images and 100 testing images. Performance evaluation is conducted based on metrics such as intersection over union, precision, recall, and latency. The results demonstrate that the anchor-based approach achieves superior accuracy with an average IoU of 0.98 but requires a longer detection time of approximately 2.33 seconds per image. Conversely, the anchor-free approach offers significantly faster processing at 0.14 seconds per detection, though with a lower average IoU of 0.78. This study concludes that while anchor-based methods excel in precision, anchor-free architectures provide alternative for time-critical applications, offering a clear reference for optimizing future face detection systems.
Bayesian Hyperparameter Optimization of Lightweight CNNs for Facial Dermatological Classification Herimanto; Intan Rumondang Sianipar; Theo Samuel Dicunawi Aritonang; Ella Tasya Marito Silaban
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 10 No 1 (2026): February 2026
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v10i1.7020

Abstract

Convolutional Neural Networks (CNNs) have been widely applied for skin condition classification. However, fair comparisons across lightweight architectures are often hindered by inconsistent hyperparameter settings. This study investigates the performance of two efficient CNN architectures, EfficientNetB3 and MobileNetV3, for facial dermatological classification across seven skin condition categories. To ensure optimal and comparable performance, Bayesian hyperparameter optimization was employed, alongside data augmentation to improve generalization. Experimental results show that EfficientNetB3 achieved the highest accuracy of 91.91%, outperforming MobileNetV3 at 90.44%. Beyond model comparison, this work highlights the novelty of applying Bayesian optimization to achieve fair benchmarking of lightweight CNNs under limited dataset conditions. The best-performing model was further deployed as a mobile application using TensorFlow Lite and Flutter, demonstrating its potential for real-world dermatological support.
Co-Authors Arie Satia Dharma Arie Satia Dharma Br Sitepu, Grace Natalia Christin Adelia Pratiwi Sihite David Largo Efendi, Syahril Ella Tasya Marito Silaban Ginting, Fastoria Gultom, Dian Jorgy Intan Rumondang Sianipar JUNITA AMALIA Lumbantobing, Samuel Adika Muhammad Zarlis Nababan, Anton Roycar Samosir, Kevin Siahaan, Ranty Deviana Simbolon, Iustisia Natalia Simbolon, Ivanston Simbolon, Niar Fujita Theo Samuel Dicunawi Aritonang