Indonesian Journal of Electrical Engineering and Computer Science
Not a member yet
9109 research outputs found
Sort by
Efficiently tracking and recognition of human faces in real-time video stream with high accuracy and performance
Get PDFReal time tracking and recognition of human faces in video streams is a critical challenge in computer vision. Existing systems often struggle to balance accuracy and performance, particularly in dynamic environments with varying lighting conditions, occlusions, and rapid movements. High computational overhead and latency further hinder their deployment in realworld applications. These limitations underscore the need for a robust solution capable of maintaining high accuracy and real-time efficiency under diverse conditions. This research addresses these challenges by developing a deep learning-based system that efficiently tracks and recognizes human faces in real-time video streams. Proposed system integrates advanced face detection models you only look once version 5 (YOLOv5) with state-of-theart tracking algorithms, such as deep simple online and real time tracking (SORT), to ensure consistency and robustness. By leveraging graphics processing unit (GPU) acceleration, the system achieves optimal performance while minimizing latency. Multi-frame analysis techniques are incorporated to enhance accuracy in detecting and recognizing faces, even under challenging conditions such as partial occlusions and motion blur. Developed system has broad applications across multiple domains, including surveillance and security, where it can enhance real-time monitoring in crowded environments for seamless face tracking in interactive systems. By focusing on efficiency, robustness, and adaptability this work offering a scalable and high-performance solution for real-time human face tracking and recognition
Implementation of innovative deep learning techniques in smart power systems
Get PDFThe integration of deep learning techniques into smart power systems has gained significant attention due to their potential to optimize energy management, enhance grid reliability, and enable efficient utilization of renewable energy sources. This research article explores the enhanced application of deep learning techniques in smart power systems. It provides an overview of the key challenges faced by traditional power systems and presents various deep learning methodologies that can address these challenges. The results showed that the root mean square errors (RMSE) for the weekend power load forecast were 18.4 for the random forest and 18.2 for the long short-term memory (LSTM) algorithm, while 28.6 was predicted by the support vector machine (SVM) algorithm. The authors' approach provides the most accurate forecast (15.7). After being validated using real-world load data, this technique provides reliable power load predictions even when load oscillations are present. The article also discusses recent advancements, future research directions, and potential benefits of utilizing deep learning techniques in smart power systems
A smart wearable posture correcting device based on spine curvature and vibration measurement
Get PDFIn the United States, aalmost $50 billion is expended in neck pain therapy each year. Poor posture, which affects the primary tendon responsible for reproducing finished tasks on time, has previously been recognized as a major source of upper spine discomfort. The primary objective of this study is to design and develop a device that not only detects deviations in posture but also employs vibration alerts to encourage corrective actions. The methodology involves the integration of an inertial measurement unit (IMU) sensor and a Flex Sensor to measure the angle and position of the spine, enabling real-time posture assessment. Additionally, a Piezo-electric sensor is incorporated to measure the vibration of the user's spine. The device provides real-time feedback via a mobile application to help users maintain optimal posture. Data analysis involved filtering and machine learning-based classification to assess posture deviations. The system demonstrated an accuracy of 90% in classifying posture states, with an average error of 2.7° in spine curvature measurement. This research contributes to the field of wearable technology by offering an innovative solution for posture correction, emphasizing the importance of proactive interventions in fostering healthy habits
Enhancing SDN security using ensemble-based machine learning approach for DDoS attack detection
Get PDFSoftware-defined networking (SDN) is a groundbreaking technology that transforms traditional network frameworks by separating the control plane from the data plane, thereby enabling flexible and efficient network management. Despite its advantages, SDN introduces vulnerabilities, particularly distributed denial of service (DDoS) attacks. Existing studies have used single, hybrid, and ensemble machine learning (ML) techniques to address attacks, often relying on generated datasets that cannot be tested because of accessibility issues. A major contribution of this study is the creation of a novel, publicly accessible dataset, and benchmarking the proposed approach against existing public datasets to demonstrate its effectiveness. This paper proposes a novel approach that combines ensemble learning models with principal component analysis (PCA) for feature selection. The integration of ensemble learning models enhances predictive performance by leveraging multiple algorithms to improve accuracy and robustness. The results showed that the ensemble of random forests (ENRF) model achieved the highest performance across all metrics with 100% accuracy, precision, recall, and F1-score. This study provides a comprehensive solution to the limitations of existing models by offering superior performance, as evidenced by the comparative analysis, establishing this approach as the best among the evaluated models
Optimizing stress resistance in MEMS inertial sensors through material and thickness variations
Get PDFStress on the micro-electromechnical system (MEMS) sensors significantly decreases sensor accuracy. Thermomechanical stresses induced by the packaging assembly process and external loads during operation induce a shift in the output signal (offset) of MEMS sensors. To achieve high precision in accelerometers, gyroscopes, and other MEMS devices, it is crucial to employ advanced modeling and simulation techniques to mitigate stress-induced offset drift. Therefore, this paper aims to explore and simulate stress on inertial sensors by designing a gyroscope tuning fork with a perforated proof mass to reduce the damping effect. Our findings provide insights for decreasing stress by varying the material and thickness of the inertial sensor. The least stress was obtained from an inertial silicon sensor with 5 and 20 mm thicknesses
Advanced cervical cancer classification: enhancing pap smear images with hybrid PMD filter-CLAHE
Get PDFCervical cancer remains a significant health problem, especially in developing countries. Early detection is critical for effective treatment. Convolutional neural networks (CNN) have shown promise in automated cervical cancer screening, but their performance depends on pap smear image quality. This study investigates the impact of various image preprocessing techniques on CNN performance for cervical cancer classification using the SIPaKMeD dataset. Three preprocessing techniques were evaluated: PeronaMalik diffusion (PMD) filter for noise reduction, contrast-limited adaptive histogram equalization (CLAHE) for image contrast enhancement, and the proposed hybrid PMD filter-CLAHE approach. The enhanced image datasets were evaluated on pretrained models, such as ResNet-34, ResNet-50, SqueezeNet-1.0, MobileNet-V2, EfficientNet-B0, EfficientNet-B1, DenseNet121, and DenseNet-201. The results show that hybrid preprocessing PMD filter-CLAHE can improve the pap smear image quality and CNN architecture performance compared to the original images. The maximum metric improvements are 13.62% for accuracy, 10.04% for precision, 13.08% for recall, and 14.34% for F1-score. The proposed hybrid PMD filter-CLAHE technique offers a new perspective in improving cervical cancer classification performance using CNN architectures
Textual and numerical data fusion for depression detection: a machine learning framework
Get PDFDepression, a widespread mood disorder, significantly affects global mental health. To mitigate the risk of recurrence, early detection is crucial. This study explores socioeconomic factors contributing to depression and proposes a novel machine learning (ML)-based framework for its detection. We develop a tailored questionnaire to collect textual and numerical data, followed by rigorous feature selection using methods like backward removal and Pearson’s chi-squared test. A variety of ML algorithms, including random forest (RF), support vector machine (SVM), and logistic regression (LR), are employed to create a predictive classifier. The RF model achieves the highest accuracy of 96.85%, highlighting its effectiveness in identifying depression risk factors. This research advances depression detection by integrating socioeconomic analysis with ML, offering a robust tool for enhancing predictive accuracy and enabling proactive mental health interventions
Devising the m-learning framework for enhancing students' confidence through expert consensus
Get PDFPast research has shown the relationship between self-regulated learning (SRL) and academic success. Self-regulated learners will monitor their learning, reflect on what they have learnt, adjust their learning strategies accordingly, and repeat this entire process throughout their learning. The ability to perform SRL will require the individual to have the belief and confidence in his/her capacity to succeed and accomplish the tasks. Therefore, this study aims to devise a mobile learning (m-learning) framework for enhancing the students’ confidence. To achieve this, the Fuzzy Delphi method was used to validate the proposed framework where the survey questionnaire was distributed to 21 experts who are the experts in their respective fields for their consensus to be obtained. Consensus showed that “assessment data” can indicate the students’ confidence when they attempt the assessment. Experts opined that “goal expectation,” and “viewed lessons, chapters, or syllabus” exert the most influence on the students’ confidence when they attempt their assessment. There was strong consensus from experts that “data security” is the most important element in the system infrastructure, and the “text mining technique” element can be used to evaluate the students’ confidence
An optimized architecture for real-time fraud detection in big data systems, ecosystems, and environments
Get PDFThe exponential growth of data in recent years has created significant challenges in fraud detection. Fraudulent activities are increasingly widespread across sectors, such as banking, web networks, health insurance, and telecommunications. This trend highlights a growing need for big data technologies such as Hadoop, Spark, Storm, and HBase to enable real-time detection and analysis of data fraud. This study aims to enhance understanding of the fraud classifications and their spread in various sectors. Fraud detection involves analyzing data and developing machine learning (ML) models or traditional rule-based systems to identify abnormal activities as they occur. The analysis in this paper examines both the advantages and limitations of these solutions, particularly regarding scalability and performance. This paper evaluates the methods and big data tools used in fraud detection and prevention through a comprehensive literature review, emphasizing the implementation challenges. This review discusses existing solutions, operational environments, and the ML algorithms and traditional rules employed. The main objective of this study is to address these challenges by proposing an innovative architecture that equips organizations with the latest knowledge and methodologies in big data technologies for real-time fraud detection and prevention
The impact of coordinator failures on the performance of Zigbee networks in various topologies
Get PDFZigbee, a key technology in the field of wireless networks for the Internet of Things, plays a significant role in the development of modern wireless network technologies. In this study, the analysis of coordinator failures in ZigBee networks with different topologies (“star”, “tree”, “mesh”) was carried out using the OPNET Modeler software tool. The problems related to the reliability and efficiency of systems using Zigbee technology are considered. Simulation of successive coordinator failures allowed us to compare the performance of topologies, revealing that the tree topology provides high traffic speed and bandwidth, but suffers from significant packet loss and delays. In turn, the star topology demonstrates minimal latency and high speed, and the mesh topology has better reliability with less packet loss, but the lowest speed and bandwidth. The findings emphasize the importance of choosing the optimal topology to ensure the efficiency and reliability of Zigbee networks in a volatile environment and increased load