Kinetik: Game Technology, Information System, Computer Network, Computing, Electronics, and Control
Not a member yet
424 research outputs found
Sort by
YOLOv9-Assisted Vision System for Health Assessment in Poultry Using Deep Neural Networks
No full textPoultry farming represents one of the fastest-growing sectors in global food production, yet disease outbreaks, high mortality, and labor shortages continue to threaten its sustainability. Conventional health monitoring methods based on visual inspection are time-consuming, subjective, and inadequate for early anomaly detection. In response, computer vision and deep learning have emerged as transformative tools for livestock management. While prior implementations of the YOLO object detection family, such as YOLOv5 and YOLOv8, have achieved notable success, their performance often deteriorates in dense flocks, low-light conditions, and occlusion-prone environments. This study introduces a YOLOv9-assisted vision framework tailored for poultry health assessment in commercial farm settings. The system integrates smart cameras with edge computing to enable real-time detection of behavioral and physiological anomalies without dependence on high-bandwidth or cloud-based resources. A dataset of 903 annotated poultry images, categorized into healthy and sick classes, was employed for model development. The trained model achieved 88.7% precision, 97% recall, an F1-score of 0.82, and a [email protected] of 0.88, demonstrating robustness under variable illumination, bird occlusion, and high-density environments. Comparative evaluation confirmed that YOLOv9 provides a superior balance of accuracy, generalization, and computational efficiency relative to YOLOv8–YOLOv11, supporting practical deployment on edge devices. Limitations include the binary scope of health classification and reliance on a single dataset. Future directions involve extending the framework to multi-class disease recognition, cross-dataset validation, behavior-based temporal modeling, and multimodal fusion, thereby advancing predictive analytics and welfare-oriented poultry farming
Speed Synchronization of Multi-Conveyor System Using Bidirectional Interaction Topologies
No full textLong production lines composed of multiple stand-alone mode controllers often face challenges when speed synchronization is required, as each setpoint must be manually adjusted one by one. The issue can be addressed by designating one conveyor as the leader, while the others operate as followers that continuously adjust their speeds to match the leader. The main objective of this study is to develop a multi-conveyor leader–follower system based on distributed cooperative control, allowing all follower conveyors to maintain synchronized speeds with the designated leader unit. In this setup, the leader is equipped with the ability to command all followers to align their speeds to its own, which is governed by a fuzzy logic controller (FLC). Each follower operates in one of two modes: a stand-alone FLC mode or a synchronization mode using cooperative control. The cooperative control mechanism relies on speed information shared among neighboring conveyors, as defined by the system topology. Two types of bidirectional interaction topologies are explored in this work: The Bidirectional Coordinated Conveyor Topology (BCCT) and the Bidirectional Leader Coordinated Conveyor Topology (BLCCT). The proposed control strategy was tested on a mini multi-conveyor setup with one leader and four followers. Synchronization tests on two topologies produced RMSE values of 30.88 RPM for BCCT and 43.87 RPM for BLCCT. A brief disturbance was also applied to one follower to assess the controller’s resilience and its effect on overall system coordination. The study confirms that combining fuzzy logic with cooperative control enhances synchronization and coordination across conveyors
Comparative Evaluation of BM25–FAISS and Small-LLM–GPT in Retrieval-Augmented Generation Concept Map Assessment
No full textConcept map-based assessment is a practical approach to measure students’ conceptual understanding, but manual assessment still faces challenges such as subjectivity, inconsistency, and limited scalability. This study proposes the application of Retrieval-Augmented Generation (RAG) as an artificial intelligence-based automated assessment solution in an educational context. The objectives of this study are to compare the effectiveness of two retrieval methods, BM25 and FAISS, and to analyse the trade-off between large-scale generative models (GPT) and Small-LLM in assessing concept map propositions. This study uses a quantitative experimental approach by combining a retriever and a generator in the RAG system. Performance evaluation is carried out using the Macro-F1 and QWK metrics to measure agreement with expert judgment, and the Explanation Relevance Score (ERS) to assess explanation quality. The experimental results show that the FAISS–GPT combination achieves the best performance, with a Macro-F1 of 0.338 and a QWK of 0.146, slightly superior to the BM25–GPT combination. In contrast, the use of Small-LLM, both with BM25 and FAISS, showed lower performance with Macro-F1 values in the range of 0.167–0.221 and QWK close to zero. This finding confirms that semantic-based retrieval plays a vital role in improving the accuracy of automated assessment, while large-scale generative models are more effective in representing conceptual relationships in depth. This study contributes through a comparative analysis of retrievers and generators, and by introducing ERS as an additional metric for RAG-based automated assessment in the field of education
Weighted ANOVA and Mutual Information for Enhanced Intrusion Detection System
No full textThe rapid escalation in the sophistication of network attacks has exposed the limitations of traditional Intrusion Detection Systems (IDS). While machine learning has shown great promise in enhancing IDS performance, its success often hinges on the effectiveness of feature selection. Standard feature selection techniques, however, struggle in cybersecurity applications due to the highly imbalanced nature of network traffic datasets. In such settings, minority attack classes—though critical—are often overshadowed by majority classes, leading to reduced detection of rare intrusions. To address this challenge, we propose a hybrid feature selection framework that integrates Analysis of Variance (ANOVA) and Mutual Information (MI) with a novel class-frequency weighting mechanism. This weighting scheme adjusts the relevance score of each feature according to the distribution of classes, ensuring that features associated with rare attacks are more strongly emphasized during the selection process. We evaluate our method on the UNSW-NB15 dataset using a Support Vector Machine classifier. The results show that our approach achieves substantial gains in recall for underrepresented classes while simultaneously reducing feature dimensionality and maintaining efficiency. By improving the visibility of features tied to minority attacks, the proposed framework provides a more balanced and reliable solution for modern IDS. This contribution advances the detection of rare but impactful threats and highlights a scalable pathway for building more resilient cybersecurity defenses
Adaptive EKF-Based Ship Trajectory Estimation with Earth Curvature Modeling and Dynamic Noise Tuning
No full textAccurate position estimation is critical for the effectiveness of automated weapon and navigation systems. Standard Extended Kalman Filter (EKF) models typically adopt flat-Earth assumptions and static noise covariances, which limit their accuracy in operational environments. This study proposes an optimized EKF framework that integrates two complementary approaches. First, ship trajectories are represented in Earth-Centered Earth-Fixed (ECEF) coordinates with a WGS-84 reference to account for Earth’s curvature. Second, process (Q) and measurement (R) covariances are adaptively determined using Joint Likelihood Maximization (JLM) with logarithmic scale exploration, enabling the filter to automatically identify the most accurate configuration. Each Q/R setting is evaluated within the EKF framework using root mean square error (RMSE) derived from radar data logs. The method was tested under short-history scenarios (5 and 10 data points) within an operational range of ±15 km, reflecting conditions commonly encountered in Combat Management Systems (CMS). The results show that while coordinate transformation alone provides only marginal improvements at short ranges, the combination of curvature modelling and adaptive Q/R tuning significantly reduces RMSE, achieving average errors approaching zero with high repeatability as measured by standard deviation. This research demonstrates a novel integration of geometric and statistical optimization in EKF design and highlights its applicability to ship trajectory estimation and defence systems
Website Quality Evaluation of OKE Garden using WebQual, The Marketing Mix, and Importance-Performance Analysis
No full textThe rapid advancement of digital technologies has significantly impacted various service sectors, including the garden landscaping industry. In response to this development, OKE Garden has implemented a website-based e-commerce platform aimed at improving service accessibility and operational efficiency. This study seeks to evaluate the usability and service quality of this digital platform from the user’s perspective by adopting the Technology Acceptance Model (TAM) as the analytical framework. Within this framework, Perceived Ease of Use (PEOU) is assessed using WebQual 4.0 indicators, while Perceived Usefulness (PU) is measured through the four elements of the marketing mix, namely Product, Price, Place, and Promotion. To analyze the alignment between user expectations and actual service performance, the Importance-Performance Analysis (IPA) method was utilized. Data were obtained from 57 respondents in the Greater Jakarta area (Jabodetabek), primarily first-time users who had previously interacted with the OKE Garden website. Prior to analysis, the data underwent validity and reliability testing to ensure robustness. The findings show that users rated the importance of website attributes higher than their actual performance, indicating a gap that highlights areas requiring improvement. Several key indicators were identified—including ease of navigation, clarity of information, data security, and pricing strategy—which were categorized in Quadrant I (high importance, low performance), indicating areas that require immediate attention. Overall, the results suggest that although digital technology adoption has taken place, user acceptance remains suboptimal. Therefore, a more comprehensive enhancement of usability and service quality is necessary to meet user expectations and improve overall satisfaction
Power Quality Improvement in Micro Hydro Power Plant Based-ELC and VSI using Fuzzy-PI Controller
Full text linkThe stability of frequency and voltage in micro hydro power plants (MHPP) depends on the ability to maintain balance between active and reactive power while managing load variations. Active power is typically regulated by an Electronic Load Controller (ELC), while reactive power is managed by a Voltage Source Inverter (VSI), with the VSI specifically compensating for reactive power induced by inductive loads. This study aims to enhance the control of active and reactive power in an MHPP system under varying load conditions by improving the ELC and VSI using Fuzzy-PI controller. The Fuzzy-PI controller applied in the ELC ensures a more precise TRIAC firing angle, enabling accurate control of the ballast load to balance the active power. Similarly, Fuzzy-PI controller applied in the VSI provides precise reactive power compensation to counteract inductive load effects. The performance of the proposed Fuzzy-PI-based ELC and VSI was evaluated using a complete MHPP model simulated in Matlab. Results demonstrated that the improved ELC and VSI effectively enhanced the system performance. Specifically, the Fuzzy-PI controller enabled the ELC to achieve accurate active power balance, while the VSI delivered suitable reactive power compensation. Consequently, the system achieved improved frequency and voltage stability under load variations, leading to enhanced power quality in the MHPP
Bamboo Diameter Detection System Based on Image Processing as a Pre-Assessment for an Automated Bamboo Splitting Technology
Full text linkBamboo is recognized for its eco-friendly attributes and rapid growth, serves as a promising sustainable alternative to wood. However, the high production cost of laminated bamboo remains a major challenge due to labor-intensive processes, particularly manual splitting, which affects efficiency and labor costs. To overcome this issue, this study presents an automated bamboo diameter measurement system that leverages Canny Edge Detection and Hough Transform to ensure precise and uniform slat dimensions. A dataset of 100 bamboo images with diameters ranging from 11 - 13 cm was utilized for training and testing. The system achieved a high accuracy, with a coefficient of determination (R²) of 0.973, demonstrating strong predictive reliability. Furthermore, Bayesian Optimization was applied to fine-tune parameters, resulting in an optimized configuration for both Canny Edge Detection and Hough Transform. The proposed system reduces dependence on manual labor, thereby lowering production costs and improving overall manufacturing efficiency. Automation in the bamboo splitting process ensures consistent and precise slat dimensions, supporting scalability and enhancing the economic feasibility of laminated bamboo production. The findings of this study provide a practical and sustainable solution to optimize production, making laminated bamboo a more viable and competitive material in the industry
Improved Chaotic Image Encryption on Grayscale Colorspace Using Elliptic Curves and 3D Lorenz System
No full textDigital data, especially visual content, faces significant security challenges due to its susceptibility to eavesdropping, manipulation, and theft in the modern digital landscape. One effective solution to address these issues is the use of encryption techniques, such as image encryption algorithms, that ensure the confidentiality, integrity, and authenticity of digital visual content. This study addresses these concerns by introducing an advanced image encryption method that combines Elliptic Curve Cryptography (ECC) with the 3D Lorenz chaotic system to enhance both security and efficiency. The method employs pixel permutation, ECC-based encryption, and diffusion using pseudo-random numbers generated by the Lorenz 3D system. The results show superior performance, with an MSE of 3032 and a PSNR of 8.87 dB, as well as UACI and NPCR values of 33.34% and 99.64%, respectively, indicating strong resilience to pixel intensity changes. During testing, the approach demonstrated robustness, allowing only the correct key to decrypt images accurately, while incorrect or modified keys led to distorted outputs, ensuring encryption reliability. Future work could explore extending the method to color images, optimizing processing for larger datasets, and incorporating additional chaotic systems to further fortify encryption strength
Classification of Breast Cancer Histopathology Images with Attention-Based Multiple Instance Learning Method
No full textBreast cancer is one of the deadliest types of cancer among women worldwide. Early detection plays a crucial role in increasing the chances of successful treatment and reducing the risk of death. Various efforts have been made by both the general public and medical professionals to raise awareness, promote early screening, and ensure timely medical intervention. With advances in technology, the use of computer-based systems, particularly in the field of medical image analysis, has become increasingly important. One such application is histopathological image analysis to support the diagnostic process in breast cancer cases. Histopathological image classification has gained significant attention from researchers in recent years, and various machine learning and deep learning techniques have been applied to improve its accuracy. Convolutional Neural Networks (CNNs), as part of the deep learning framework, have shown promising results in identifying tissue patterns in histopathological images. However, despite their high accuracy, CNNs are often less interpretable, making it difficult to understand the reasoning behind their predictions—especially when dealing with subtle features such as small spots, dots, or fine lines that may be overlooked. This study addresses these limitations by proposing a method that not only classifies histopathological images with high accuracy but also enhances readability through localization techniques. The goal is to make the classification process more transparent and clinically useful. Using widely recognized datasets like BreakHIS, the proposed method achieves a classification accuracy of up to 97.50%, demonstrating its potential as a reliable tool in medical diagnostics and breast cancer research