Articles

Access the latest knowledge in applied science, electrical engineering, computer science and information technology, education, and health.

Advanced Search
Filter Icon

Filters article

Years

FAQ Arrow
Single Year
Range Year
0
0

Source Title

FAQ Arrow

Authors

FAQ Arrow

31,377 Article Results

ROVAA: Offline attendance automation using a voice–OCR-based 3-DOF robotic arm with Raspberry Pi

10.11591/ijra.v15i3.pp561-576
Rajanikanth Kashi Nagaraj , Archana Harihara Ranganatha , Surendra Hanumanthaiah Honnamachanahalli , Venu Manighatta Gopalakrishnappa
Conventional classroom attendance systems suffer from limitations in accuracy, hygiene, data privacy, and reliability in low-connectivity environments, whether they are manual, cloud-dependent, or single-modality systems. To address these gaps, this paper presents ROVAA, a low-cost, fully offline, AI-driven robotic attendance system in the classroom environment that uniquely integrates three complementary modalities: offline voice recognition, optical character recognition (OCR), and a 3- degrees of freedom (DOF) robotic arm controlled via inverse kinematics, an integration not demonstrated in prior work. The system operates on a Raspberry Pi 4 model B and employs the Vosk speech recognition model and Tesseract OCR for accurate offline processing. Audio and visual inputs are matched in real time to enable the arm to mark attendance at pre-calibrated positions on a touchscreen. Experimental validation under varied lighting and acoustic conditions yielded 96.2% speech recognition accuracy, 95.8% OCR accuracy, and 97.6% robotic arm precision, producing an overall system success rate of 92.8%, demonstrating that high reliability is achievable without cloud infrastructure. The system is designed for cost-effectiveness, data privacy, and scalability, making it suitable for resource-constrained environments such as rural schools and institutions with limited network access. It additionally serves as an educational platform for human–robot collaboration.
Volume: 15
Issue: 3
Page: 561-576
Publish at: 2026-09-01

Ethical issues in digital mental health delivery: a comprehensive scoping review of global challenges

10.11591/ijaas.v15.i3.pp911-920
Nurul Ain Ain Mohd Mohd Daud , Hapsah Md Yusof , Norazani Ahmad , Fauziah Mohd Sa’ad , Nurul Hasyimah Mat Rani , Pau Kee , Nur Shuhana Mohd Sansuddin
Digital technologies are becoming increasingly integrated into mental health counselling. While expanding the boundaries of service provision, these technologies have also raised numerous ethical challenges that urgently need clarification. This study conducts a scoping review following the standardized methodology developed by Arksey and O’Malley. It identifies five core ethical challenges facing practitioners: privacy and data security; informed consent and autonomy; service accessibility and equality; algorithmic fairness and bias mitigation; and inclusiveness and equity. Existing comparative analyses show that Western countries, including the United States (US) and the United Kingdom (UK), have mature digital mental health regulatory frameworks. In contrast, non-Western countries generally face gaps in legal protections, which have created a global ethical divide. Artificial intelligence (AI)-related risks, such as algorithmic bias and lack of transparency, further exacerbate these problems. The study calls on policymakers to improve relevant regulatory frameworks and requires practitioners to receive specialized ethical training.
Volume: 15
Issue: 3
Page: 911-920
Publish at: 2026-09-01

Improving ride comfort of semi-active suspension with magnetorheological damper by fuzzy-PID control

10.11591/ijaas.v15.i3.pp1038-1049
Zhang Kunlun , Azizan As’arry , Liucun Zhu , Abdul Aziz Hairuddin , Mohd Khair Hassan , Mohd Zarhamdy Md Zain
Semi-active suspension using magnetorheological (MR) damper has become one of the popular choices for improving vehicle ride comfort due to its adjustable damping, affordable cost, and quick response. MR damper has highly nonlinear and obvious hysteresis behavior, so its control becomes complicated. Fuzzy control combined with proportional-integral-derivative (PID) controller is used to study the vibration reduction control strategy of semi-active suspension with MR damper. PID is highly robust, but the control parameters cannot be changed during the control process. Therefore, it has good adjustment ability for specific situations, but cannot effectively adjust nonlinear systems. Fuzzy control performs fuzzy inference based on input variables and fuzzy rules constructed from expert knowledge, and continuously outputs effective control parameters to adjust nonlinear systems. The advantages of both are used to construct fuzzy-PID controller to control MR damper. A quarter-car semi-active suspension with MR damper model is established using MATLAB/Simulink for simulation analysis. Compared with passive and PID, fuzzy-PID reduces the root mean square (RMS) values of body acceleration, suspension travel, and tire deflection, reduces body vibration, and improves ride comfort.
Volume: 15
Issue: 3
Page: 1038-1049
Publish at: 2026-09-01

Flight supervision and pilot training standards: a study of international aviation regulations

10.11591/ijaas.v15.i3.pp1203-1213
Heni Puspita , Ade Gafar Abdullah , Agus Setiawan , Isma Widiaty , Ike Yuni Wulandari , Johanes Adi Prihantono , Erlian Supriyanto
This paper analyzes variations in pilot training and licensing policies across four major organizations: the International Civil Aviation Organization (ICAO), the Federal Aviation Administration (FAA), the European Union Aviation Safety Agency (EASA), and the Directorate General of Civil Aviation (DGCA) of Indonesia. The main issue here is the impact of contradictory training pathway requirements, flight hours, simulator requirements, language proficiency requirements, and instructor qualification requirements on aviation safety, license transferability, and the construction of the pilot training system in Indonesia. This study employs a qualitative research design, combining a systematic literature review with regulatory comparative analysis. The findings demonstrate that the prescriptive and competency-based model is most prevalent in EASA, where simulation and multi-crew requirements are widely applied; whereas in the FAA, the emphasis is more on total flight experience and rules related to 1,500 hours; and in the Civil Aviation Safety Regulations (CASR), there is minimal implementation but overall consistent with ICAO minimum. The paper finds that the country must progressively harmonize with modern ICAO, EASA, and FAA standards of good practice, prioritizing competency-based training (CBT), fatigue risk management, simulator capacity, and instructor professionalization to enhance safety performance and international recognition.
Volume: 15
Issue: 3
Page: 1203-1213
Publish at: 2026-09-01

Wheat leaf disease classification using vision transformer based deep learning approach

10.11591/ijaas.v15.i3.pp954-963
P. Ashwini Ashwini , D. Mythili , P. Jyothi , K. Swathi , G. Rama Krishna , M. Sowmya
Wheat crops can be affected by numerous fungal diseases like leaf rust, stripe rust, and stem rust. These diseases can significantly reduce yield and grain quality and also result in highly economic damage. Detection of diseases at an early stage is the main and most complex task for farmers due to common morphological properties like color, shape, texture, and edges. Deep learning (DL) offers a powerful solution for detecting wheat crop diseases due to its ability to accurately identify diseases, even in complex field conditions, and facilitate early intervention. By analyzing images, DL models can learn complex patterns associated with various diseases, enabling faster and more reliable diagnoses than traditional methods. In this paper use DL models followed by a vision transformer model (ViTM) to extract multilevel features to focus more on the disease-infected area on the plant with enhanced accuracy. DL models extract hierarchical features, whereas transformer models are utilized to capture global dependencies and contextual relationships within the image. In the proposed research evaluated various DL models, InceptionV3, EfficientNet-B0, visual geometry group 16 (VGG-16), and ResNet-50 with visual transformers on the wheat disease detection dataset of 3679 images and achieved accuracy rates of 98.7%, 97.6%, 93.9%, and 97.5%, respectively. The experiment proved that InceptionV3 outperforms the other models.
Volume: 15
Issue: 3
Page: 954-963
Publish at: 2026-09-01

A deep learning-based system for coral reef image segmentation using YOLOv8 with EfficientNet-B0 in Indonesian waters

10.11591/ijeecs.v43.i2.pp628-639
Raden Sutiadi , Sparisoma Viridi , Giyanto Giyanto , Andarta F. Khoir , Rizkie S. Utama , Elsa D. Aulia , Tri A. Hadi , Ludi Parwadani Aji
Manual coral point count with excel extension (CPCe) analysis requires approximately 4–6 hours to process one 50-image station, limiting the scale of coral reef monitoring. This study presents an artificial intelligence-based workflow using YOLOv8 with an EfficientNet-B0 backbone to automate benthic cover estimation. A total of 8,147 underwater images from 151 transects across 39 Indonesian coral reef stations were annotated into eleven benthic categories for model training, while evaluation was conducted using the 2021 Derawan Islands dataset. During validation, YOLOv8 achieved an average mAP@0.5 of 0.58, recall of 0.79, and an average absolute percentage cover error of 9.8% compared with CPCe. The model processed each image in 3.13 seconds, equivalent to 156.47 seconds per 50-image station, representing a 92–138× speedup over manual CPCe analysis. These results show that the proposed workflow can support scalable and near-real-time coral reef monitoring across Indonesia.
Volume: 43
Issue: 2
Page: 628-639
Publish at: 2026-08-01

Feature-guided transformer approach for detecting distributed denial of service attacks

10.11591/ijai.v15.i4.pp3712-3721
Lokeshwaran Kanagaraj , Raguraman Purushothaman , Sathya Subramanian , Durga Devi Saravanan , Komal Kumar Napa , Cornelius Karunakaran , Billa Manindhar
Distributed denial of service (DDoS) attacks continues to pose serious risks to modern networks, with their growing intensity making early detection both critical and challenging. Conventional machine learning (ML) models often struggle with the nonlinear and highly dynamic nature of attack traffic, which motivates the use of advanced architectures. In this study investigate a transformer-based classifier for DDoS detection on the CIC-DDoS2019 dataset. The workflow included preprocessing, feature scaling, and domain-guided feature selection. Logistic regression (LR) was employed as a baseline, achieving 92.1% accuracy and F1-score of 0.90, thereby revealing the limitations of linear models. The transformer, after hyperparameter tuning and 5-fold cross-validation, reached an average accuracy of 99.95% with precision, recall, and F1-scores all above 99.9%. The model demonstrated stable convergence and generalization across folds. These results highlight the strength of attention mechanisms in capturing feature dependencies, while also pointing to future directions such as real-time deployment, explainability, and resilience to zero-day attacks.
Volume: 15
Issue: 4
Page: 3712-3721
Publish at: 2026-08-01

Experimental evaluation of an artificial intelligence system for automated mango classification

10.11591/ijai.v15.i4.pp3919-3933
Surasit Phokha , Wiriya Dangton , Viroch Sukontanakarn , Phisit Srinoi
This study proposes the design and implementation of an intelligent mango sorting system capable of classifying mangoes into three categories: unripe, ripe, and rotten. The system integrates a programmable logic controller (PLC) to operate a conveyor mechanism, with user interaction facilitated through a human-machine interface (HMI) touchscreen panel. Image acquisition is performed using a universal serial bus (USB) webcam, while image processing and classification are handled by the CiRA CORE software utilizing artificial intelligence (AI) techniques. The classification results are transmitted to an Arduino microcontroller, which controls pneumatic actuators responsible for the physical sorting process. Experimental results demonstrate that the system can accurately classify mangoes with an overall success rate of 90%, indicating its potential for practical application in automated agricultural product sorting.
Volume: 15
Issue: 4
Page: 3919-3933
Publish at: 2026-08-01

Integration of artificial intelligence technology in learning evaluation using the ADDIE method

10.11591/ijai.v15.i4.pp3068-3080
Adip Wahyudi , Slamet Arifin , Mukhammad Solikhin , Alfi Sahrina , Feri Fahrian Maulana , Mohammad Firzon Ainur Roziqin , Maria Emerita Indraningrum Shrestha
Artificial intelligence offers significant potential to enhance learning evaluation, yet implementing fair and objective assessments remains a challenge. This study developed an artificial intelligence-based learning evaluation system to improve assessment inclusivity and efficiency. Employing the research and development method with the analysis, design, development, implementation, and evaluation (ADDIE) model, the system features a virtual assistant chatbot and an automated grading tool artificial intelligence assessment. The product underwent validation by media and material experts, followed by field trials with 20 secondary school teachers. The results demonstrated high feasibility, with validation scores of 96% by user interface/user experience (UI/UX) and 91.2% by material experts, and an implementation feasibility score of 94.4%. The system successfully assisted teachers in reducing administrative burdens and providing personalized feedback based on student learning styles. Future developments should expand subject coverage and support diverse assessment formats.
Volume: 15
Issue: 4
Page: 3068-3080
Publish at: 2026-08-01

Comparison of machine learning algorithms to identify muscular atrophy and muscular dystrophy

10.11591/ijai.v15.i4.pp3505-3517
Ganga Bhavani Billa , Venkateswara Rao Peddada
Muscular dystrophy is a genetic disorder characterized by progressive muscle weakness and degeneration, while muscular atrophy involves the loss of muscle tissue due to pathological conditions. Accurate differentiation between these neuromuscular disorders is critical for effective clinical diagnosis and treatment planning. This study presents a machine learning– based framework for classifying muscular dystrophy and muscular atrophy using surface electromyography (sEMG) signals. sEMG data were acquired and processed to extract informative features with low computational complexity. Several machine learning classifiers several techniques were used and assessed, such as random forest (RF), support vector machine (SVM), gradient boosting (GB), and extreme gradient boosting (XGBoost). To further enhance classification performance, a novel ensemble learning approach combining multiple machine learning models was proposed. Experimental results exhibit that the suggested ensemble model completes significantly higher grouping accurateness and robustness compared to individual classifiers and existing methods reported in the literature. The results of this investigation indicate that ensemble-based machine learning techniques integrated with sEMG analysis provide a reliable and non-invasive diagnostic aid, with strong potential for early detection and improved clinical management of neuromuscular disorders.
Volume: 15
Issue: 4
Page: 3505-3517
Publish at: 2026-08-01

A machine learning framework for skin cancer classification using texture descriptors

10.11591/ijai.v15.i4.pp3556-3567
Shikha Malik , Vaibhav V. Dixit
Skin cancer remains a critical health and economic concern worldwide. Timely and accurate diagnosis is crucial for improving mortality rate of patients. Although automated machine learning (ML) models assist doctors in clinical assessment of skin cancer from dermoscopic images, their performance often suffers from extreme class imbalance, as benign image samples greatly exceed malignant ones. This leads to false detection and delayed diagnosis of skin cancer. To overcome this issue, the study proposes an efficient and lightweight geometric transformation (GT)–augmented support vector machine (SVM) framework for early diagnosis of skin cancer. It effectively addresses the class imbalance issues present in the datasets and improves the detection of positive cases. The novel pipeline integrates preprocessing, morphology preserving GT, rotation-invariant texture feature extraction, feature validation, and feature scaling for performing binary classification using an optimized SVM framework. This framework provides balanced and accurate lesion classification even if image samples are insufficient. Experimental results have successfully achieved a true positive rate (TPR) of 93% on PH2 and 81.1% on International Skin Imaging Collaboration 2016 (ISIC-2016) dataset, which is better than conventional ML models. These findings prove that proposed GT-SVM framework is a lightweight, interpretable, and computationally efficient approach for early skin cancer diagnosis. Future developments shall explore multi-class lesion classification, validation across diverse clinical datasets, and hybrid feature fusion.
Volume: 15
Issue: 4
Page: 3556-3567
Publish at: 2026-08-01

Resource optimization via identification of articles with aspects of priority and high utility

10.11591/ijai.v15.i4.pp3240-3251
Garima Srivastava , Vaishali Singh , Sachin Kumar
Owing to the linear economic growth, the manufacturing of consumer products has increased manifold, products in diversified packaging and formats are now available in abundance, sometimes even with less or no requirement, resulting in losses. Addressing the gap that exists between productivity and requirements can minimize losses and unnecessary burden on manufacturing units. To minimize losses, a hybrid algorithm is proposed, using the advantages offered by classification and high utility patterns to identify the dominant aspects of articles with a high probability of sale. Articles are identified in a two-step process, identification of samples with context as prominent parameter by vanilla feedforward neural network (VNN) as a first step. The second step comprises the determination of utility pattern mining in articles using faster high-utility itemset miner (FHN), sentiment score obtained provides the utility pattern of the product. Proposed hybrid VNN + FHN, along with convolutional neural network (CNN), long short-term memory (LSTM), and transformer-based prediction model (TPM) were used for assessing the utility-driven product analysis. The hybrid VNN + FHN proposed displays the best adaptability by extracting 158 patterns with a utility of 162.88 units. The algorithm outperforms CNN in utility, LSTM and TPM in speed, and CNN in pattern count, making it a better choice for resource optimization.
Volume: 15
Issue: 4
Page: 3240-3251
Publish at: 2026-08-01

WVisionBERT-VL: a multimodal model architecture for toxicity classification on social media platforms using large language models

10.11591/ijai.v15.i4.pp3365-3375
Witta Listiya Ningrum , Achmad Benny Mutiara , Diana Ikasari
The increasing prevalence of toxic content on social media, conveyed through text, images, and videos, poses significant challenges for automated content moderation systems. Although prior studies have reported promising results in unimodal and bimodal settings, they often fail to capture implicit and contextual toxicity emerging from interactions across multiple modalities, particularly in non-English environments. This paper proposes WVisionBERT-VL, an end-to-end multimodal framework for toxicity detection that integrates text, image, and video modalities within a unified architecture. The proposed model incorporates modality-specific encoders, bidirectional multi-head cross attention (BMHCA) for cross-modal synchronization, and an adaptive fusion gate to dynamically balance modality contributions. A balanced multimodal dataset is constructed from social media platforms, including X, Instagram, and TikTok, and refined using a model-based labeling strategy with limited human-in-the-loop validation. Experimental results on a custom Indonesian dataset demonstrate strong in-domain performance, achieving an accuracy of 94.12%, a macro-F1 of 0.9407, and a receiver operating characteristic - area under the curve (ROC-AUC) of 0.9721, with robustness further validated through five-fold cross-validation. Cross-dataset evaluation highlights challenges related to domain shift, underscoring the need for future research on robust and domain-adaptive multimodal toxicity detection.
Volume: 15
Issue: 4
Page: 3365-3375
Publish at: 2026-08-01

Image-based estimation of surface roughness in Al-7075 drilling

10.11591/ijai.v15.i4.pp3492-3504
Shilpa M. Karegoudra , Vamsidhar Yendapalli
Surface roughness is an important quality parameter that affects the performance, durability, and reliability of machined components. Measuring the internal surface roughness of drilled holes using conventional contact-based methods is often difficult. Accessibility of the internal surface is complicated due to the diameter of the drilled hole and the need to interrupt the machining process. To overcome these limitations, the proposed work offers a non-contact method for estimating the surface roughness of drilled Al-7075 using image-based analysis. Drilled surfaces are machined using different speeds and feed rates. High-resolution images of the drilled surface are captured using a custom-built image-capturing setup. Texture and frequency features were extracted using gray-level co-occurrence matrix (GLCM), discrete Fourier transform (DFT), and discrete wavelet transform (DWT) techniques. Surface arithmetic average roughness (Ra) was measured using these extracted features. The estimated roughness parameters were then validated by comparing them against roughness parameters obtained by means of the contact stylus technique. According to the experimental results, the accuracy of the wavelet technique is higher, with mean absolute errors of 0.32 μm compared to those obtained using GLCM and DFT techniques. The findings demonstrate that the proposed image-based framework is a reliable and practical solution for non destructive surface roughness prediction.
Volume: 15
Issue: 4
Page: 3492-3504
Publish at: 2026-08-01

E-TEXTLOC: efficient text localization and extraction in real-world video scenes

10.11591/ijai.v15.i4.pp3537-3545
Dayananda Kodala Jayaram , Puttegowda Devegowda
Localization of an accurate text is quite a complicated issue, especially when attempting to extract from a complex video. It is mainly due to the limitation of resources, dynamic background, and text variability. There is various artificial intelligence based methods adopting machine learning for addressing such issues encounter lower positional accuracy and incur maximized computational cost. Therefore, the proposed system introduces efficient text localization and extraction for comprehensive positional accuracy in complex videos (E-TEXTLOC). Different from conventional approaches, E-TEXTLOC facilitates sampling of video frames while MobileNetV2 is deployed towards faster localization of text. The outcome of recognized text is further refined by a verifier module, which provides a self-supervised response. Assessed on the YouTube video dataset, the proposed model accomplishes 98.2% accuracy with 29.6 ms towards generating analytical outcomes. It means the proposed model accomplishes 6-10% accuracy enhancement with a 40-50% reduction of speed in contrast to the existing system. The implications of the proposed study can be stated towards surveillance system, autonomous vehicles, and assistive devices that works in real-time.
Volume: 15
Issue: 4
Page: 3537-3545
Publish at: 2026-08-01
1 2 3
...
2091 2092
Show 36 of 2092

Discover Our Library

Embark on a journey through our expansive collection of articles and let curiosity lead your path to innovation.

Explore Now
Library 3D Ilustration