Abstract:The management of academic data at SMA Negeri 1 Tampahan previously operated semi-digitally, where grade management relied on isolated spreadsheet files and attendance was documented manually on paper journals. This unintegrated…
tegrated mechanism created operational redundancies, increased the risk of data inconsistency, and substantially slowed down the E-Rapor generation process due to repetitive data cross-checking by homeroom teachers. This study aims to develop a web-based Integrated Academic Information System that consolidates student attendance and subject grading data to streamline the data preparation process for E-Rapor entry. The system was developed using the Waterfall model, comprising requirement analysis, system design, coding, testing, and maintenance. The software architecture was constructed with PHP utilizing the Laravel framework, Bootstrap for responsive front-end styling, and MySQL for relational database management. Black Box Testing was performed across four user roles: Administrator, Subject Teachers, Homeroom Teachers, and Students. The functional testing yielded a 100% success rate across all critical modules, including automated grading aggregation, attendance recapitulation, and consolidated academic report generation. The implementation provides accurate, pre-processed datasets ready for E-Rapor processing, minimizing administrative workloads and mitigating human error in final academic reporting.
Abstract:This study aims to formulate a preservation strategy for Kauman Village in Semarang as a sustainable cultural and religious tourism destination through contextual corridor design. As an embryo of urban growth, Kauman currently…
rently faces physical degradation and identity erosion driven by uncontrolled modernization. The research methodology adopts Roger Trancik's integrated urban design theory, incorporating figure/ground (mass and void), linkage (visual-structural connectivity), and place (socio-cultural meaning of space) analyses, which are further synthesized through a SWOT matrix. The results formulate the "Wonderful of Java" design concept, a physical transformation that reintegrates traditional Joglo architecture and Islamic symbolism into the main corridor gates. This layout also offers tactical solutions, including dedicated parking facilities and traditional culinary zoning, to mitigate circulation congestion. The study concludes that spatial interventions rooted in local philosophy effectively preserve the dual "santri" and "Jawa" identity amidst capitalistic shifts, while successfully revitalizing the area's historical character as a religious public space.
Abstract:Abstract: Falls represent a leading cause of injury-related mortality among elderly individuals, demanding reliable automated detection deployable in resource-constrained environments. Existing vision-based approaches predominantly…
edominantly employ machine learning classifiers requiring large labelled datasets, substantial training pipelines, and GPU-class hardware, impeding rapid clinical deployment. This paper presents an explainable, fully rule-based real-time fall detection system implemented in Python using MediaPipe Pose, designed to operate on commodity edge devices without any training dataset. The system comprises three tightly integrated layers: (1) a mathematical formulation layer computing vector-based torso angle, distance-ratio body normalization, and frame-to-frame CoM velocity; (2) a temporal modeling layer combining a deterministic four-state Finite State Machine with a sliding-window motion energy analysis; and (3) a robustness evaluation across three camera viewing angles, two lighting conditions, and three fall types. Experimental validation yields a sensitivity of 91.8%, specificity of 95.8%, and F1-score of 93.1%, with a mean end-to-end detection latency of 140.1 ms on Raspberry Pi 4—satisfying a 200 ms real-time constraint while consuming only 72.5% mean CPU and 284 MB RAM. The rule-based architecture provides full decision auditability without black-box models, aligning with emerging XAI requirements for medical device regulation.
Keywords: Periodic Table of Elements; Chemistry Teaching Materials; Spiritual Values; Discovery Learning.
Abstrak: Kondisi Jatuh merupakan penyebab utama kematian akibat cedera di kalangan lansia, sehingga membutuhkan deteksi otomatis yang andal dan dapat diterapkan di lingkungan dengan keterbatasan sumber daya. Pendekatan berbasis visi yang ada sebagian besar menggunakan pengklasifikasi pembelajaran mesin yang membutuhkan kumpulan data berlabel besar, alur pelatihan yang substansial, dan perangkat keras kelas GPU, yang menghambat penerapan klinis yang cepat. Makalah ini menyajikan sistem deteksi jatuh waktu nyata yang dapat dijelaskan dan sepenuhnya berbasis aturan yang diimplementasikan dalam Python menggunakan MediaPipe Pose, yang dirancang untuk beroperasi pada perangkat edge komoditas tanpa kumpulan data pelatihan apa pun. Sistem ini terdiri dari tiga lapisan yang terintegrasi erat: (1) lapisan formulasi matematika yang menghitung sudut tubuh berbasis vektor, normalisasi tubuh rasio jarak, dan kecepatan CoM antar frame; (2) lapisan pemodelan temporal yang menggabungkan Mesin Keadaan Terbatas empat keadaan deterministik dengan analisis energi gerak jendela geser; dan (3) evaluasi ketahanan di tiga sudut pandang kamera, dua kondisi pencahayaan, dan tiga jenis jatuh. Validasi eksperimental menghasilkan sensitivitas 91,8%, spesifisitas 95,8%, dan skor F1 93,1%, dengan latensi deteksi ujung-ke-ujung rata-rata 140,1 ms pada Raspberry Pi 4—memenuhi batasan waktu nyata 200 ms sambil hanya mengonsumsi 72,5% CPU rata-rata dan 284 MB RAM. Arsitektur berbasis aturan memberikan kemampuan audit keputusan penuh tanpa model kotak hitam, selaras dengan persyaratan XAI yang muncul untuk regulasi perangkat medis.
Kata Kunci: Deteksi jatuh, MediaPipe Pose, komputasi tepi, sistem berbasis aturan, kecerdasan buatan dengan keadaan terbatas, AI yang dapat dijelaskan, perawatan lansia, pemantauan real-time
Abstract:Abstract: Hydrometeorological disasters frequently damage electrical and telecommunication infrastructures, disrupting the coordination of Search and Rescue (SAR) operations. Conventional communication devices such as Handy…
ndy Talkies (HT) are limited to voice communication and do not support real-time multimedia data transmission. This study aims to develop a portable emergency communication system based on Low Earth Orbit (LEO) Starlink satellite internet integrated with a solar power system as an independent energy source to support communication during post-disaster emergency response. The research employs the Research and Development (R&D) approach using the Design Science Research (DSR) paradigm, consisting of requirement identification, system architecture design, prototype development, laboratory testing, and field simulation. System performance is evaluated based on Quality of Service (QoS) parameters, including latency, throughput, connection stability, power consumption, operating duration, and system portability. The prototype is designed to weigh less than 15 kg, require less than 10 minutes for deployment, and operate independently for 6–8 hours using a 300–500 W solar power system. The proposed system is expected to provide reliable internet connectivity for real-time transmission of audio, video, images, and GPS coordinates, thereby improving the effectiveness of SAR coordination in disaster-affected areas.
Keywords: Emergency Communication; Starlink; Solar power; Search and Rescue; Satellite Network.
Abstrak: Bencana hidrometeorologi sering menyebabkan kerusakan infrastruktur listrik dan telekomunikasi sehingga menghambat koordinasi operasi Search and Rescue (SAR). Keterbatasan perangkat komunikasi konvensional seperti Handy Talky (HT) mengakibatkan pertukaran informasi hanya terbatas pada komunikasi suara tanpa dukungan data multimedia secara real-time. Penelitian ini bertujuan mengembangkan sistem komunikasi darurat portabel berbasis internet satelit Low Earth Orbit (LEO) Starlink yang terintegrasi dengan panel surya sebagai sumber energi mandiri untuk mendukung komunikasi pada kondisi pascabencana. Metode penelitian menggunakan pendekatan Research and Development (R&D) dengan paradigma Design Science Research (DSR) yang meliputi identifikasi kebutuhan, perancangan arsitektur sistem, pengembangan prototipe, pengujian laboratorium, dan simulasi lapangan. Evaluasi dilakukan terhadap parameter Quality of Service (QoS) berupa latensi, throughput, stabilitas koneksi, konsumsi daya, durasi operasi, serta portabilitas sistem. Prototipe dirancang berbobot kurang dari 15 kg, waktu pemasangan kurang dari 10 menit, dan mampu beroperasi mandiri selama 6–8 jam menggunakan panel surya berkapasitas 300–500 W. Sistem yang dikembangkan diharapkan mampu menyediakan konektivitas internet yang stabil untuk mendukung komunikasi audio, video, citra, dan koordinat GPS secara real-time, sehingga meningkatkan efektivitas koordinasi operasi SAR di wilayah terdampak bencana..
Kata kunci: Komunikasi darurat; Starlink; Panel Surya; Search and Rescue; Jaringan Satelit.
Abstract:Abstract: Traffic sign detection and recognition (TSDR) constitute fundamental components of intelligent transportation systems (ITS) and autonomous driving technologies. Accurate and real-time recognition of traffic signs…
ns is essential for improving road safety, driver assistance systems, and autonomous vehicle navigation. However, challenges such as illumination variations, occlusions, weather conditions, small object sizes, and complex backgrounds significantly affect detection performance. This study proposes an Attention-Enhanced YOLOv8 framework for real-time traffic sign detection and recognition by integrating the Convolutional Block Attention Module (CBAM) into the YOLOv8 architecture. The proposed model aims to improve feature extraction capabilities and detection accuracy, particularly for small and partially occluded traffic signs. Experiments were conducted using the German Traffic Sign Detection Benchmark (GTSDB) dataset and additional Indonesian traffic sign datasets. The proposed model achieved a mean Average Precision (mAP@0.5) of 97.4%, precision of 96.8%, recall of 95.9%, and inference speed of 74 FPS, outperforming conventional YOLOv8 and other state-of-the-art methods. Experimental results demonstrate that the integration of attention mechanisms significantly enhances detection performance while maintaining real-time capabilities suitable for intelligent transportation applications.
Keywords: Traffic Sign Detection, YOLOv8, Attention Mechanism, Intelligent Transportation Systems, Deep Learning.
.
Abstrak: Deteksi dan pengenalan rambu lalu lintas (TSDR) merupakan komponen mendasar dari sistem transportasi cerdas (ITS) dan teknologi kemudi otonom. Pengenalan rambu lalu lintas yang akurat dan berlangsung secara real-time sangat penting untuk meningkatkan keselamatan jalan, sistem bantuan pengemudi, dan navigasi kendaraan otonom. Namun, berbagai tantangan seperti variasi pencahayaan, oklusi, kondisi cuaca, ukuran objek yang kecil, dan latar belakang yang kompleks sangat memengaruhi kinerja deteksi. Penelitian ini mengusulkan kerangka kerja YOLOv8 yang diperkuat dengan mekanisme attention (Attention-Enhanced YOLOv8) untuk deteksi dan pengenalan rambu lalu lintas secara real-time dengan mengintegrasikan Convolutional Block Attention Module (CBAM) ke dalam arsitektur YOLOv8. Model yang diusulkan bertujuan untuk meningkatkan kemampuan ekstraksi fitur dan akurasi deteksi, khususnya untuk rambu lalu lintas yang berukuran kecil dan mengalami oklusi sebagian. Eksperimen dilakukan menggunakan dataset German Traffic Sign Detection Benchmark (GTSDB) dan dataset tambahan berupa rambu lalu lintas Indonesia. Model yang diusulkan mencapai nilai mean Average Precision (mAP@0.5) sebesar 97,4%, presisi 96,8%, recall 95,9%, dan kecepatan inferensi 74 FPS, yang mengungguli YOLOv8 konvensional serta metode-metode mutakhir (state-of-the-art) lainnya. Hasil eksperimen menunjukkan bahwa integrasi mekanisme attention secara signifikan meningkatkan kinerja deteksi sekaligus mempertahankan kemampuan real-time yang sesuai untuk aplikasi transportasi cerdas.
Kata kunci: Deteksi Rambu Lalu Lintas, YOLOv8, Mekanisme Attention, Sistem Transportasi Cerdas, Deep Learning.
Abstract:Abstract: The development of transformer architecture has brought significant changes in Natural Language Processing (NLP) research, particularly in semantic parsing tasks. This study aims to analyze the development of transformer-based…
ransformer-based semantic parsing through a Systematic Literature Review (SLR) approach. The research method uses the PRISMA guidelines with a process of identification, selection, and analysis of articles from the Scopus, ScienceDirect, IEEE Xplore, Springer, and Google Scholar databases for the period 2019–2026. The results of the study show that transformer models such as BERT, T5, GPT, and transformer encoder–decoder have become the dominant approaches in modern semantic parsing because they are able to improve contextual understanding, semantic representation, and reasoning more effectively than previous methods. The most widely used datasets include Spider, WikiSQL, GeoQuery, ATIS, and Overnight, with the main evaluation being Execution Accuracy and Exact Match Accuracy. This study also identifies various challenges, such as limited labeled data, high computational costs, low model interpretability, the risk of hallucinations, and limitations in low-resource languages. This research contributes to a literature review, identifying research gaps, and recommending future research directions for transformer-based semantic parsing.
Keywords: BERT, Natural Language Processing, Semantic Parsing, Systematic Literature Review, Transformer, T5, Text-to-SQL.
Abstrak: Perkembangan arsitektur transformer telah membawa perubahan signifikan dalam penelitian Natural Language Processing (NLP), khususnya pada tugas semantic parsing. Penelitian ini bertujuan untuk menganalisis perkembangan semantic parsing berbasis transformer melalui pendekatan Systematic Literature Review (SLR). Metode penelitian menggunakan pedoman PRISMA dengan proses identifikasi, seleksi, dan analisis artikel dari database Scopus, ScienceDirect, IEEE Xplore, Springer, dan Google Scholar pada periode 2019–2026. Hasil kajian menunjukkan bahwa model transformer seperti BERT, T5, GPT, dan encoder–decoder transformer menjadi pendekatan dominan dalam semantic parsing modern karena mampu meningkatkan contextual understanding, semantic representation, dan reasoning secara lebih efektif dibandingkan metode sebelumnya. Dataset yang paling banyak digunakan meliputi Spider, WikiSQL, GeoQuery, ATIS, dan Overnight, dengan metrik evaluasi utama berupa Execution Accuracy dan Exact Match Accuracy. Penelitian ini juga mengidentifikasi berbagai tantangan, seperti keterbatasan data berlabel, tingginya biaya komputasi, rendahnya interpretabilitas model, risiko hallucination, serta keterbatasan pada low-resource language. Penelitian ini berkontribusi dalam memberikan pemetaan literatur, identifikasi research gap, dan rekomendasi arah penelitian semantic parsing berbasis transformer pada masa mendatang.
Kata Kunci: BERT, Natural Language Processing, Semantic Parsing, Systematic Literature Review, Transformer, T5, Text-to-SQL.
Abstract:The number of rail passengers on the Kisaran–Medan route varies over time due to a number of variables, including local economic situations, national holidays, peak travel seasons, and current transportation regulations.…
s. The administration of Kisaran Station finds it challenging to manage train fleets, organize travel timetables, and enhance service quality due to this unpredictability. Using historical data from 2019 to 2025, this study intends to apply an Artificial Neural Network (ANN) with the Backpropagation technique to forecast the number of train passengers at Kisaran Station. With a learning rate of 0.25 and a sigmoid binary activation function, the system was constructed using a 7-5-1 network architecture with seven input neurons (yearly data from 2019–2025), five hidden layer neurons, and one output neuron. The data was split into 70% training data (January–July) and 30% testing data (August–December) after being adjusted to the interval 0.1–0.9 using the Min-Max technique. At epoch 822, the training process reached a convergence. For most months, including March (predicted: 647,271; actual: 642,870 passengers) and April (predicted: 999,728; actual: 996,320 passengers), the model was able to produce predictions that were close to actual values, according to testing results. However, there were differences in some months with seasonal spikes. It is anticipated that PT Kereta Api Indonesia and the regional government would use the created prediction system as a decision-support tool to plan transportation services and enhance the general quality of railway services.
Abstract:Extreme weather in coastal areas such as Kota Padang is influenced by complex topography and local atmospheric dynamics, making daily rainfall prediction a critical challenge for disaster mitigation and sectoral planning.…
. This study aims to develop a daily rainfall prediction model using a Deep Neural Network (DNN) and evaluate its performance. The dataset consists of daily temperature, humidity, and air pressure data collected from the Automatic Weather Station (AWS) of BMKG Maritime Teluk Bayur from January 2020 to December 2024, comprising 1,823 samples after preprocessing. The research methodology follows the CRISP-DM framework, which includes six main phases. The proposed DNN architecture contains three hidden layers (128, 64, and 32 neurons) with ReLU activation and dropout regularization, enhanced by lag and rolling average features to capture temporal patterns. The model achieved an MAE of 1.79, RMSE of 17.31, and R² of 0.8972, indicating strong predictive performance. The model was deployed through a Streamlit-based interactive dashboard
Abstract:Abstract: The conventional and manual personnel budget execution reporting (Laplakgar) architecture in public defense management is highly susceptible to prolonged processing cycles and recurring data reconciliation anomalies.…
alies. This study aims to develop, model, and technically articulate an Artificial Intelligence (AI)-driven integrated information system architecture—conceptualized as the SIAPGAR application model—to automate hierarchical data aggregation and enforce real-time, layered data verification at Military Institution X. A Research and Development (R&D) methodology utilizing a system prototyping approach was applied in this study. The system's logical and data boundaries were systematically mapped through structural Data Flow Diagrams (DFDs) at the Context Level (Level 0) and Functional Level (Level 1). Structural success parameters and data integrity were evaluated based on the System Quality and Information Quality dimensions derived from the DeLone and McLean Information Systems Success Model, and validated through technical expert judgment involving database administrators and financial software operators. The architectural engineering of the SIAPGAR model successfully automates three error-prone manual data processing bottlenecks. The functional DFD model demonstrates that the integrated AI validation layer operates as an automated gatekeeper, systematically identifying data variances between allocated budgets and actual personnel expenditures before cross-level aggregation occurs. This automated algorithmic intervention results in a projected 129% increase in workflow efficiency and effectively reduces human-caused data discrepancies to near zero. This study presents an empirical blueprint for an AI-integrated public finance system architecture. The development of SIAPGAR demonstrates that automated operational verification at the lower level successfully feeds accurate data into a real-time analytical dashboard essential for strategic defense decision-making.
Keywords: Information systems, Data Flow Diagrams, Artificial Intelligence, SIAPGAR, Budget Execution.
Abstrak: Arsitektur pelaporan pelaksanaan anggaran (Laplakgar) personel yang konvensional dan manual dalam manajemen pertahanan publik sangat rentan terhadap siklus pemrosesan yang berkepanjangan dan anomali rekonsiliasi data yang berulang. Studi ini bertujuan untuk mengembangkan, memodelkan, dan mengartikulasikan secara teknis arsitektur sistem informasi terintegrasi yang digerakkan oleh Kecerdasan Buatan (AI)—dikonseptualisasikan sebagai model aplikasi SIAPGAR—untuk mengotomatisasi agregasi data hierarkis dan menegakkan verifikasi data berlapis secara real-time di Institusi Militer X. Metodologi Research and Development (R&D) yang memanfaatkan pendekatan prototyping sistem diterapkan dalam penelitian ini. Batasan logis dan data sistem dipetakan secara sistematis melalui Data Flow Diagram (DFD) struktural pada Tingkat Konteks (Level 0) dan Tingkat Fungsional (Level 1). Parameter keberhasilan struktural dan integritas data dievaluasi berdasarkan dimensi Kualitas Sistem dan Kualitas Informasi yang diturunkan dari DeLone and McLean Information Systems Success Model, serta divalidasi melalui penilaian ahli (expert judgment) teknis yang melibatkan administrator pangkalan data dan operator perangkat lunak keuangan. Rekayasa arsitektur dari model SIAPGAR berhasil mengotomatisasi tiga hambatan pemrosesan data manual yang rentan terhadap kesalahan. Model DFD fungsional mendemonstrasikan bahwa lapisan validasi AI yang terintegrasi beroperasi sebagai penjaga gerbang (gatekeeper) otomatis, secara sistematis mengidentifikasi varians data antara pagu keuangan yang dialokasikan dengan pengeluaran personel aktual sebelum agregasi lintas tingkat terjadi. Intervensi algoritmik otomatis ini menghasilkan proyeksi peningkatan efisiensi alur kerja sebesar 129% dan secara efektif menekan perbedaan data akibat faktor manusia hingga mendekati nol. Studi ini menyajikan cetak biru empiris untuk arsitektur sistem keuangan publik yang diintegrasikan dengan AI. Pengembangan SIAPGAR membuktikan bahwa verifikasi operasional otomatis pada tingkat bawah berhasil memasok data yang akurat ke dalam dasbor analitik real-time yang esensial untuk pengambilan keputusan pertahanan yang strategis.
Kata Kunci: Sistem informasi, Data Flow Diagram, Kecerdasan Buatan, SIAPGAR, Pelaksanaan anggaran.
Abstract:Abstract: Oil palm leaf diseases can significantly reduce harvest productivity. Manual identification is considered less effective because it requires a long time and carries the risk of misidentification. This study develops…
elops a leaf image detection system using a CNN with the MobileNetV2 architecture, implemented on both web and mobile platforms. A dataset of 1,200 images from Kaggle and field observations covers four classes: leaf spot, anthracnose, yellow stripe, and Other. The research stages include data collection, image preprocessing, dataset splitting, CNN model training using the MobileNetV2 architecture, model testing, performance evaluation, and system implementation on web and mobile platforms. Evaluation was conducted using confusion matrix, accuracy, precision, recall, and F1-score parameters. Based on testing results, the model achieved 93.33% accuracy, 93.71% precision, 92.92% recall, and 93.31% F1-score. The developed system was successfully implemented on web and mobile platforms with features such as image upload, automatic classification, detection history, and disease management recommendations. With this system, users are expected to perform early detection of oil palm leaf diseases more quickly, practically, and accurately.
Keywords: CNN, Deep Learning, Image Classification, MobileNetV2, Oil Palm Leaf Disease.
Abstrak: Penyakit daun kelapa sawit dapat menurunkan produktivitas panen secara signifikan. Identifikasi manual dinilai kurang efektif karena membutuhkan waktu lama dan berpotensi salah identifikasi. Penelitian ini membangun sistem deteksi citra daun menggunakan CNN arsitektur MobileNetV2 berbasis web dan mobile. Dataset 1.200 citra dari Kaggle dan observasi lapangan mencakup empat kelas: bercak daun, antraknosa, garis kuning, dan Other. Tahapan penelitian meliputi pengumpulan data, preprocessing citra, pembagian dataset, pelatihan model CNN menggunakan arsitektur MobileNetV2, pengujian model, evaluasi performa, serta implementasi sistem berbasis web dan mobile. Proses evaluasi dilakukan menggunakan parameter confusion matrix, accuracy, precision, recall, dan F1-score. Berdasarkan hasil pengujian, model memperoleh akurasi 93,33%, precision 93,71%, recall 92,92%, dan F1-score 93,31%. Sistem yang dibangun berhasil diimplementasikan pada platform web dan mobile dengan fitur unggah citra, proses klasifikasi otomatis, riwayat deteksi, serta rekomendasi penanganan penyakit. Dengan adanya sistem ini, pengguna diharapkan dapat melakukan deteksi dini penyakit daun kelapa sawit secara lebih cepat, praktis, dan akurat.
Kata Kunci: CNN, Deep Learning, Klasifikasi Citra, MobileNetV2, Penyakit Daun Sawit.