Claim Missing Document
Check
Articles

Found 27 Documents
Search

Optimizing Coral Fish Detection: Faster R-CNN, SSD MobileNet, YOLOv5 Comparison Santoso, Syifa Afnani; Jaya, Indra; Priandana, Karlisa
IJCCS (Indonesian Journal of Computing and Cybernetics Systems) Vol 18, No 2 (2024): April
Publisher : IndoCEISS in colaboration with Universitas Gadjah Mada, Indonesia.

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22146/ijccs.95011

Abstract

This study underscores the critical role of accurate Chaetodontidae fish abundance observations, particularly in assessing coral reef health. By integrating deep learning algorithms (Faster R-CNN, SSD-MobileNet, and YOLOv5) into Autonomous Underwater Vehicles (AUVs), the research aims to expedite fish identification in aquatic environments. Evaluating the algorithms, YOLOv5 emerges with the highest accuracy, followed by Faster R-CNN and SSD-MobileNet. Despite this, SSD-MobileNet showcases superior computational speed with a mean average precision (mAP) of around 92.21% and a framerate of about 1.24 fps. Furthermore, employing the Coral USB Accelerator enhances computational speed on the Raspberry Pi 4, enabling real-time detection capabilities. This study incorporates centroid tracking, facilitating accurate counting by assigning unique IDs to identified objects per class. Ultimately, the real-time implementation of the system achieves 87.18% accuracy and 87.54% precision at 30 fps, empowering AUVs to conduct real-time fish detection and tracking, thereby significantly contributing to underwater research and conservation efforts.
Acoustic Sediment Classification Using High-Frequency (400 kHz) Multibeam Data in Pari Water of Seribu Island, Indonesia Handoko, Dadang; Manik, Henry Munandar; Hestirianoto, Totok; Priandana, Karlisa; Hasan, Rozaimi Che
ILMU KELAUTAN: Indonesian Journal of Marine Sciences Vol 30, No 1 (2025): Ilmu Kelautan
Publisher : Marine Science Department Diponegoro University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.14710/ik.ijms.30.1.135-144

Abstract

Seafloor classification is essential for understanding sediment distribution, marine habitat characteristics, and resource management. Therefore, this study aimed to classify seafloor sediment in the Pari water, Indonesia using high-frequency (400 kHz) backscatter data obtained through the Multibeam Echosounder T-50P. The Angular Range Analysis (ARA) method was applied to analyze backscatter intensity variations across different incidence angles, to enhance the accuracy of sediment classification in this shallow marine environment. Data acquisition was collected using the T-50P, which captured high-resolution acoustic signals from varying angles to generate angular response curves. Analysis was conducted in the curves were then analyzed to differentiate sediment types, with ground-truth sediment samples collected to validate classification outcomes. The result showed that backscatter intensity mosaic had an intensity range of -27 dB to -37.5 dB. Applying ARA enabled the identification of 12 sediment classes, including sandy silt, coarse silt, and clayey sand. Sediment distribution maps, generated via FMGT and visualized with ArcGIS, indicated a predominance of fine-grained sediments. The FMGT-based classification tended to prioritize finer sediment categories, likely due to the acoustic limitations in detecting granular details. Conversely, the in-situ analysis of 15 sediment samples revealed medium sand as the predominant sediment type, accompanied by smaller proportions of coarse sand and coral fragments. The discrepancies between the in-situ sampling and FMGT results were primarily due to the operational frequency of the MBES system, which limits the acoustic signal's penetration to the surface of the seabed. This highlights the importance of in-situ sampling to complement acoustic data, especially in accurately seabed characterization. 
Long Short Term Memory-Based Marine Data Prediction with Pearson Correlation Mukhlis, Mukhlis; Jaya, Indra; Nurdiati, Sri; Priandana, Karlisa; Hermadi, Irman
PIKSEL : Penelitian Ilmu Komputer Sistem Embedded and Logic Vol. 13 No. 1 (2025): Maret 2025
Publisher : LPPM Universitas Islam 45 Bekasi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33558/piksel.v13i1.10731

Abstract

Marine data prediction plays a vital role in supporting decision-making in the field of marine environment and resources. However, the complexity of marine data, which is nonlinear and dynamic, is a significant challenge in producing accurate predictions. This study aims to explore the role of Long Short-Term Memory (LSTM) models in computer systems to predict marine data, focusing on Pearson Correlation analysis. The methods applied include collecting historical marine data, implementing LSTM models for prediction, and evaluating performance using metrics such as Mean Absolute Error (MAE). In addition, Pearson Correlation analysis is used to understand the relationship between variables in marine data. The results show that the LSTM model is able to produce predictions with a low error rate with a composition of training data and testing data of 80:20, resulting in Sea Surface Temperature (SST) = 0.0053, Sea Surface Salinity (SSS) = 0.0026, sea Surface Height (SSH) = 0.0061 and CHL-a = 0.0002 and shows a significant relationship between variables through Multivariate correlation analysis. This research contributes to the development of marine data-based prediction systems and provides implications for the world of marine resource research and management.
An Intelligent Food Recommendation System for Dine-in Customers with Non-Communicable Diseases History Harry Imantho; Kudang Boro Seminar; Evy Damayanthi; Nugraha Edhi Suyatma; Karlisa Priandana; Bonang Waspadadi Ligar; Annisa Utami Seminar
Jurnal Keteknikan Pertanian Vol. 12 No. 1 (2024): Jurnal Keteknikan Pertanian
Publisher : PERTETA

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.19028/jtep.012.1.140-152

Abstract

The rising prevalence of diet-related diseases necessitates a focus on individual food selection to enhance nutrition intake and promote overall health. This study introduces a novel food recommender system utilizing artificial intelligence, specifically a genetic algorithm (GA), to intelligently match diverse nutritional needs with available food items. The research incorporates machine learning methodologies, such as collaborative and content-based filtering, to develop a recommendation model. Data from a commercial restaurant, Nutrisurvey, and the Indonesian food composition list inform the nutritional analysis of five menu items. Consumer variability, considering factors like sex, body mass index, medical conditions, and physical activity, are integrated into the GA framework for personalized food pattern matching. The presented results demonstrate the efficacy of the proposed model in offering tailored food recommendations for consumers with non-communicable diseases (NCDs), such as diabetes, hypertension, and heart disease. The multi-objective optimization technique employed in the system ensures a balance between nutritional adequacy and individual preferences. The presented GA-based approach holds promise for promoting healthier food choices tailored to individual needs, contributing to the broader goal of fostering a sustainable and personalized food system.
Multi-Platform Detection of Melon Leaf Abnormalities Using AVGHEQ and YOLOv7 Ishak, Sahrial Ihsani; Priandana, Karlisa; Wahjuni, Sri
JOIN (Jurnal Online Informatika) Vol 10 No 1 (2025)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v10i1.1441

Abstract

This research develops a multiplatform system for detecting abnormalities in melon leaves, integrating an Internet of Things (IoT) approach using Jetson Nano, a Streamlit-based website, and a mobile application for real-time monitoring. The system employs preprocessing with Average Histogram Equalization (AVGHEQ) to enhance image quality, followed by modeling with the YOLOv7 algorithm on a dataset of 469 training images and 52 test images, validated through 5-fold cross-validation. The model achieved a mean Average Precision (mAP) of 84% with an inference detection time of 4.5 milliseconds. Implementation on Jetson Nano resulted in a 25% increase in CPU usage (from 25% to 50%) and a 20% increase in RAM usage (from 70% to 90%). By combining these platforms and leveraging robust data preprocessing and modeling techniques, the system provides an accessible, efficient, and scalable solution for agricultural monitoring, enabling farmers to address plant health issues promptly and effectively.
Biological constraint in digital data encoding: A DNA based approach for image representation Muttaqin, Muhammad Rafi; Herdiyeni, Yeni; Buono, Agus; Priandana, Karlisa; Siregar, Iskandar Zulkarnaen; Kusuma, Wisnu Ananta
International Journal of Advances in Intelligent Informatics Vol 11, No 3 (2025): August 2025
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26555/ijain.v11i3.1747

Abstract

Digital data encoding is crucial for communication and data storage, but conventional techniques, such as ASCII and binary coding, have drawbacks in terms of processing speed and storage capacity. A potential substitute with parallel processing and high-capacity storage is DNA-based data encoding. The goal of this research is to develop a digital data encoding technique based on DNA, while considering biological constraints such as homopolymer and GC-content. The process involves converting image pixel values into binary format, followed by encoding into DNA sequences, ensuring they meet biological constraints. The validity of the resulting DNA sequences is assessed through transcription and translation processes. Additionally, Multiple Sequence Alignment analysis is conducted to compare the similarities between the encoded DNA sequences. The results indicate that the DNA sequences from MNIST images share similar characteristics, reflected in the phylogenetic tree's close clustering. Multiple Sequence Alignment analysis shows that biological constraints successfully preserved the core visual features, allowing accurate clustering. However, this method also faces drawbacks, particularly in the reduction of visual information and sensitivity to changes in image intensity. Despite these challenges, DNA-based encoding shows potential for digital image representation. Further development, particularly the integration of deep learning, could lead to more efficient, secure, and sustainable data storage systems, especially for image data.
Hybrid convolutional vision transformer for extrusion-based 3D food-printing defect classification Mawardi, Cholid; Buono, Agus; Priandana, Karlisa; Herianto, Herianto
IAES International Journal of Artificial Intelligence (IJ-AI) Vol 14, No 4: August 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijai.v14.i4.pp3311-3323

Abstract

Deep learning is generally used to perform remote monitoring of three-dimensional (3D) printing results, including extrusion-based 3D food printing. One of the widely used deep learning algorithms for defect detection in 3D printing is the convolutional neural network (CNN). However, the process requires high computational costs and a large dataset. This research proposes the Con4ViT model, a hybrid model that combines the strengths of vision transformer with the inherent feature extraction capabilities of CNN. The locally extracted features in the CNN were merged using the transformers’ global features with four transformer encoder blocks. The proposed model has a smaller number of parameters compared to other lightweight pre-trained deep learning models such as VGG16, VGG19, EfficientNetB2, InceptionV3, and ResNet50. Thus, the proposed model is simplified. Simulations were conducted to classify defect and non-defect images obtained from the printing results of a developed extrusion-based 3D food printing device. Simulation results showed that the model produced an accuracy of 95.43%, higher than the state-of-the-art techniques, i.e., VGG16, VGG19, MobileNetV2, EfficientNetB2, InceptionV3, and ResNet50, with accuracies of 77.88, 86.30, 82.95, 90.87, 84.62, and 93.83%, respectively. This research shows that the proposed Con4ViT model can be used for 3D food printing defect detection with high accuracy.