Indonesian Journal of Electrical Engineering and Computer Science
Not a member yet
9109 research outputs found
Sort by
Efficient object detection for augmented reality based english learning with YOLOv8 optimization
Get PDFThis study develops a mobile-based augmented reality (AR) application with machine learning for elementary school students to enhance basic English vocabulary learning. The application integrates an optimized YOLOv8 object detection model, designed to recognize 20 common classroom objects in real-time. The model optimization involves replacing standard Conv layers with GhostConv and the C2f block with the C2fCIB block that has significantly improved computational efficiency. Evaluation results show the optimized model reduces the parameters by 22.003% and decreases the file size from 6.2 MB to 4.9 MB. The model performance improved by achieving precision of 83.7%, recall of 73.5% and a mean Average Precision (mAP) of 81.4%. The model was integrated into the Unity platform via the Barracuda library, enabling real-time detection and interactive display of 3D objects. This aplication also complete with English text, translations, example sentences also audio pronunciation. 3D objects representing classroom vocabulary were specifically created to support AR-based learning. Performance testing on a Samsung A14 showed an improved frame rate of 6–12 FPS compared to the original model’s 5–10 FPS. These results demonstrate that the optimized YOLO model effectively integrates with AR technology, creating a more interactive and enjoyable vocabulary learning experience
Optimization of single electron transistor based digital logic design
Get PDFThis paper addresses the challenge of high-power consumption and delay in conventional complementary metal-oxide-semiconductor (CMOS) circuits, particularly in the design of digital logic gates. The objective is to develop a hybrid CMOS-single-electron transistor (SET) model that reduces power consumption while maintaining acceptable performance in terms of delay. The proposed model leverages coulomb oscillation in SETs to create a changeable transconductance area, which significantly reduces energy usage. Simulation results demonstrates that the hybrid CMOS-SET circuits achieve up to 30% lower power dissipation compared to traditional CMOS designs, although a slight increase in delay is observed in complex gates like the OR gate. The novelty of this work lies in its use of coulomb oscillation for dynamic transconductance control, providing an innovative approach to balancing power efficiency and speed in nano-scale digital circuits. This makes the proposed model a promising candidate for future low-power, high-performance integrated circuits
Simulation of reactive flow over a parabolic vertical plate using MATLAB
Get PDFThis article examines how fluid flows around an infinitely large, parabolic-shaped vertical plate, which is heated at an exponentially accelerating rate and undergoes a chemical reaction with the fluid. The plate’s temperature increases at an exponential rate, adding complexity to the heat transfer process. Additionally, the fluid undergoes a chemical reaction in this environment, impacting both the flow and concentration of chemical species. The article includes graphs that show how different parameters such as the rate of temperature increase, strength of thermal radiation, and reaction rate, effect the flow, heat, and concentration profiles. This graphical analysis provides a visual understanding of how each parameter influences the behavior of the fluid
Enhance big data security based on HDFS using the hybrid approach
Get PDFHadoop has emerged as a prominent open-source framework for the storage, management, and processing of extensive big data through its distributed file system, known as Hadoop distributed file system (HDFS). This widespread adoption can be attributed to its capacity to provide reliable, scalable, and cost-effective solutions for managing large datasets across diverse sectors, including finance, healthcare, and social media. Nevertheless, as the significance and scale of big data applications continue to expand, the challenge of ensuring the security and safeguarding of sensitive data within Hadoop has become increasingly critical. In this study, the authors introduce a novel strategy aimed at bolstering data security within the Hadoop storage framework. This approach specifically employs a hybrid encryption technique that leverages the advantages of both advanced encryption standard (AES) and data encryption standard (DES) algorithms, whereby files are encrypted in HDFS and subsequently decrypted during the map task. To assess the efficacy of this method, the authors performed experiments with various file sizes, benchmarking the outcomes against other established security measures
A curvilinear-based approach for sign-to-text conversion of Kannada deaf sign language
Get PDFThis research addresses the challenge of translating Kannada sign language into text to improve communication for the deaf community. Existing methods, primarily shape-based approaches, often fail to accurately imprisonment the complexity of hand gestures, leading to reduced translation accuracy. This study proposes a curvilinear-based approach that leverages peak curvature features and contour evolution techniques to overcome these limitations. This method enhances the recognition and interpretation of sign language gestures while reducing processing overhead. Experimental results demonstrate that the proposed system significantly outperforms traditional methods, achieving higher precision and recall rates. The enhanced system provides a reliable solution for improving accessibility and communication for the deaf community. This research represents a significant step toward developing more inclusive digital communication tools, with future work focused on real-time processing and extending the system to other regional sign languages
Predictive modelling of osteoporosis and effect of BMI on the risk of fracture in femur bone using COMSOL Multiphysics: a computational modelling approach
Get PDFThis study explores the intricate relationship between osteoporosis, body mass index (BMI), and the risk of femur fractures using computational modeling. Osteoporosis is a silent metabolic disorder that depletes bone density and structure, significantly increasing the risk of fractures, particularly in weight-bearing bones such as the femur. To analyze the impact of mechanical stress on osteoporotic bones, COMSOL Multiphysics was utilized to simulate stress distribution in a femur under varying BMI conditions, providing valuable insights into how BMI influences bone health and fracture risk. A three-dimensional (3D) femur model was designed using computer-aided design (CAD) software, with specific material properties assigned for both healthy and osteoporotic bones. Finite element analysis was conducted by applying different load conditions, representing body weight, on the femur head. The results highlighted stress distribution and deformation patterns, identifying regions most prone to fracture. The findings demonstrate that while higher BMI typically correlates with increased bone density, it also leads to greater deformation in osteoporotic bones under stress, emphasizing the complex interplay between BMI and bone strength. These insights underscore BMI’s critical role in fracture risk management. Future research should incorporate advanced fracture mechanics models and clinical data to enhance predictive accuracy and develop targeted strategies for fracture prevention in osteoporotic patients
Credit card fraud detection using CNN and LSTM
Get PDFCredit card fraud is an evolving problem with the fraudsters developing new technologies to perform fraud. Fraudsters have found diverse ways to make a fraud transaction to the card holder. Thus, detecting suspicious behavior of a card is critical for preventing fraudulent transactions to happen. Artificial intelligence techniques, in particular deep learning algorithms can tackle these credit card fraud attacks by identifying patterns that predict transactions as fraud or legitimate. One-dimensional convolutional neural network (1D CNN) and long short-term memory (LSTM) both performs well on the sequential data especially on transactions data, yet there are not many studies done on combining these two algorithms to make an effective fraud detection approach. However, the dataset is highly imbalanced containing only 492 fraud transaction out of two lacs transactions. In this experimental study, firstly datasets will get prepared by using different sampling techniques along with their hybrid techniques secondly, observing the performance of individual CNN and LSTM on the datasets, finally on those datasets in which CNN and LSTM are performing well, by implementing ensemble on those data. The performance of the ensembles is observed using the performance metrics namely accuracy, F1-score, precision and recall. In the proposed experimental study, getting the F1-score of 99.96% and 99.89% in ensemble: early fusion and ensemble: late fusion respectively
CGDE-YOLOv5n: a real-time safety helmet-wearing detection algorithm
Get PDFDue to numerous parameters and calculations, existing safety helmetwearing detection models are challenging to deploy on embedded devices. Therefore, this paper proposed a you only look once (YOLO) v5n-based lightweight detection algorithm called CGDE-YOLOv5n to address the shortcomings in the following areas: (i) the YOLOv5n algorithm was selected to minimize the model’s parameters and calculations, reducing the hardware cost. (ii) The convolutional block attention module (CBAM) was integrated into the backbone to enhance the network’s feature extraction capability. (iii) The neck was improved using the efficient re-parameterized generalized feature pyramid network (efficient RepGFPN) to enhance the multi-scale object detection capability. (iv) The C3 module was improved using the deformable ConvNets v2 (DCNv2) module to enhance the network’s adaptability to geometric changes of objects. (v) The complete intersection over union (CIoU) loss was replaced with focal-efficient IoU (focal-EIoU) loss to reduce the missed detection rate. Experimental results demonstrated that the customized gradient descent estimation (CGDE)- YOLOv5n achieved a mean average precision (mAP) 50 of 89.5% and recall of 84%, which is 1% and 0.8% higher than the YOLOv5n. In particular, the recall of workers not wearing safety helmets increased by 1.7%. Furthermore, the improved model achieved a detection speed of 68.5 frames per second (FPS), meeting the real-time requirements
Blockchain-based handle-research data sharing: a blockchain-based handle system to enhance the privacy and security of research data sharing
Get PDFThe increasing demand for secure, persistent and interoperable research data (RD) sharing makes traditional systems vulnerable. All research objects should be findable, accessible, interoperable and reusable (FAIR) for machines and people. This paper proposes a novel framework called blockchain-based handle- RD sharing (BHRDS), which integrates the handle system for persistent identifiers (PIDs) with a smart contract for access control and mirror-specific encryption, BLAKE2-based hashing for identity binding and irregularity detection. The system utilizes swarm, a decentralized storage layer, for off-chain data storage while storing only credential metadata and access conditions on-chain. The framework enables secure identity data management, and verifiable credential distribution across multiple mirror sites. We conducted experiments under growing user numbers (10 to 10,000), different encryption key strengths (AES 128, 192, and 256 bits), and blockchain load conditions. Results show that BHRDS achieves high irregularity detection rates (above 97%) and maintains low response times even at scale. In all the test instances, the system performed accurately, demonstrating that BHRDS offers a decentralized data access model that is scalable and aligned with the FAIR principle, making it suitable for next-generation scientific and institutional data sharing.
Forecasting industrial electricity demand using hybrid optimization methods
Get PDFThis study presents a hybrid machine learning framework for forecasting industrial electricity consumption by comparing backpropagation neural networks (BPNN) with models enhanced through metaheuristic optimization algorithms. Using 32 years of annual data from APEC economies, the research addresses rising electricity demand driven by economic and infrastructural development. A key limitation in traditional models— underfitting due to complex data patterns—is addressed via feature selection, which identifies the most relevant variables and reduces model complexity. Five metaheuristic algorithms—cuckoo search (CS), differential evolution (DE), harmony search (HS), particle swarm optimization (PSO), and teaching–learning-based optimization (TLBO)—are applied to optimize both feature selection and BPNN training. The proposed approach improves forecasting accuracy by handling noisy inputs and capturing the nonlinear relationships common in energy datasets. Among the tested methods, TLBO consistently delivers superior accuracy and robustness across most evaluated countries. The findings contribute an effective and adaptable forecasting model with significant implications for long-term energy planning and policy development