Abstract:Abstract: The advancement of digital technology has made it easier to create, process, and distribute files—using 317 files from the dataset https://www.kaggle.com/datasets/axon data/selfie-and-official-id-photo-dataset-18k…
t-18k images?select=metadata_image.csv has also introduced new challenges, such as the increasing practice of digital file manipulation that is difficult to detect visually. Therefore, an intelligent digital forensics system that can automatically and accurately detect file authenticity is required. This study aims to develop an intelligent digital forensics system for detecting file manipulation by leveraging metadata analysis and the Random Forest classification method. The methods used include extracting metadata from digital files—such as time information, device details, and processing history—followed by analysis to identify patterns of inconsistency that indicate manipulation. This data is then used as features in the classification process using the Random Forest algorithm to distinguish between original and manipulated files. The results of this study are expected to show that the use of metadata analysis combined with the Random Forest algorithm can improve accuracy in detecting digital file manipulation compared to conventional methods. The resulting system is expected to provide an effective, efficient, and integrated solution to support digital forensic investigations, Based on the test results, the system demonstrated good performance with an accuracy rate of 94%.
Keywords: Digital Forensics;File Manipulation;Metadata Analysis;Random Forest;Classification;Machine Learning
Abstrak:Perkembangan teknologi digital telah meningkatkan kemudahan dalam pembuatan, pengolahan,dan distribusi file sebanyak 317 file, sumber datasets https:// www.kaggle.com/datasets/axondata/selfie-and-official-id-photo-dataset-18k-images?select =metadata_image.csv, namun juga menimbulkan tantangan baru berupa meningkatnya praktik manipulasi file digital yang sulit dideteksi secara kasat mata. Oleh karena itu, diperlukan suatu sistem forensik digital yang cerdas dan mampu mendeteksi keaslian file secara otomatis dan akurat. Penelitian ini bertujuan untuk mengembangkan sistem forensik digital cerdas untuk deteksi manipulasi file dengan memanfaatkan analisis metadata dan metode klasifikasi Random Forest. Metode yang digunakan meliputi proses ekstraksi metadata dari file digital, seperti informasi waktu, perangkat, dan riwayat pengolahan, kemudian dilakukan analisis untuk menemukan pola ketidaksesuaian yang mengindikasikan adanya manipulasi. Selanjutnya, data tersebut digunakan sebagai fitur dalam proses klasifikasi menggunakan algoritma Random Forest untuk membedakan antara file asli dan file yang telah dimanipulasi. Hasil dari penelitian ini diharapkan menunjukkan bahwa penggunaan analisis metadata yang dikombinasikan dengan algoritma Random Forest mampu meningkatkan akurasi dalam mendeteksi manipulasi file digital dibandingkan metode konvensional. Sistem yang dihasilkan dapat memberikan solusi yang efektif, efisien, dan terintegrasi dalam mendukung proses investigasi forensik digital, Berdasarkan hasil pengujian, sistem menunjukkan performa yang baik dengan tingkat akurasi sebesar 94%.
Kata Kunci: Forensik Digital, Manipulasi File, Metadata, Random Forest, Klasifikasi, Machine Learning.
Abstract:Abstract: Manual meeting documentation risks losing important information due to cognitive fatigue. Although automated summarization models have evolved, integrated end-to-end systems for Indonesian spoken language remain…
n highly limited. This study aims to design and evaluate an end-to-end automated meeting summarization architecture that directly integrates Automatic Speech Recognition (ASR) via OpenAI Whisper for transcription and the IndoT5 language model for abstractive summarization. IndoT5 was fine-tuned using a dataset of 486 Indonesian spoken language transcript pairs. Testing was conducted on a CPU infrastructure using MP4, MP3, and WAV formats. Results show the optimal fine-tuning configuration significantly improved accuracy, achieving ROUGE-1 (0.4167), ROUGE-2 (0.1973), and ROUGE-L (0.2701) scores. Computationally, the system achieved a Real-Time Factor below 1, processing data faster than the actual recording duration. Conclusively, integrating Whisper and IndoT5 shows potential in producing coherent meeting summaries with lightweight computational overhead, making it viable for local infrastructure implementation to ensure data privacy.
Keywords: abstractive summarization; ASR; end-to-end pipeline; IndoT5; real-time factor
Abstrak: Dokumentasi rapat manual rentan menghilangkan informasi penting akibat keterbatasan kognitif. Meskipun model peringkas otomatis telah berkembang, implementasi sistem terintegrasi (end-to-end) khusus percakapan lisan berbahasa Indonesia masih sangat terbatas. Penelitian ini bertujuan merancang dan mengevaluasi arsitektur peringkas rapat otomatis end-to-end yang mengintegrasikan langsung Automatic Speech Recognition (ASR) melalui OpenAI Whisper untuk transkripsi dan model bahasa IndoT5 untuk peringkasan abstraktif. Adaptasi domain dilakukan melalui fine-tuning IndoT5 menggunakan 486 pasang dataset transkrip lisan berbahasa Indonesia. Pengujian pada infrastruktur CPU menggunakan format MP4, MP3, dan WAV. Hasil pengujian menunjukkan konfigurasi fine-tuning optimal berhasil meningkatkan akurasi, dengan skor ROUGE-1 (0,4167), ROUGE-2 (0,1973), dan ROUGE-L (0,2701). Sistem mendemonstrasikan efisiensi komputasi dengan nilai Real-Time Factor di bawah 1, mengindikasikan waktu pemrosesan lebih cepat dari durasi rekaman asli. Kesimpulannya, integrasi Whisper dan IndoT5 menunjukkan potensi dalam menghasilkan ringkasan yang koheren dengan beban komputasi ringan, sehingga layak diimplementasikan pada infrastruktur lokal organisasi untuk menjaga privasi data.
Kata kunci: ASR; end-to-end pipeline; IndoT5; peringkasan abstraktif; real-time factor
Abstract:Abstract: The development of information and communication technology has driven digital transformation in various sectors, including agriculture. One of the technologies widely applied is the Internet of Things (IoT), which…
hich enables devices to be interconnected through the internet to perform monitoring and data exchange in real time. In hydroponic cultivation, water quality stability is a crucial factor that affects plant growth. The main parameters that need to be monitored include water acidity level (pH), water temperature, and dissolved nutrient concentration measured using Total Dissolved Solids (TDS). However, water quality monitoring is still commonly conducted manually, making it less efficient and potentially causing delays in detecting changes in water conditions. This study aims to design an IoT-based water quality monitoring system for hydroponic cultivation using an ESP32 microcontroller integrated with pH, temperature, and TDS sensors. The collected data are sent to a server and displayed on a web dashboard in real time, equipped with automatic notification features. The proposed system is expected to improve monitoring efficiency and increase the productivity of hydroponic cultivation
Keywords: esp32; hydroponics; internet of things; real-time monitoring; water quality.
Abstrak: Perkembangan teknologi informasi dan komunikasi mendorong transformasi digital dalam berbagai sektor, termasuk pertanian. Salah satu teknologi yang banyak diterapkan adalah Internet of Things (IoT) yang memungkinkan perangkat saling terhubung melalui jaringan internet untuk melakukan pemantauan dan pertukaran data secara real-time. Pada budidaya hidroponik, kestabilan kualitas air menjadi faktor penting yang mempengaruhi pertumbuhan tanaman. Parameter utama yang perlu diperhatikan meliputi tingkat keasaman air (pH), suhu air, serta konsentrasi nutrisi terlarut yang diukur menggunakan Total Dissolved Solids (TDS). Namun, pemantauan kualitas air masih banyak dilakukan secara manual sehingga kurang efisien dan berpotensi menimbulkan keterlambatan dalam mendeteksi perubahan kondisi air. Penelitian ini bertujuan merancang sistem pemantauan kualitas air hidroponik berbasis IoT menggunakan mikrokontroler ESP32 yang terintegrasi dengan sensor pH, suhu, dan TDS. Data dikirim ke server dan ditampilkan pada web dashboard secara real-time serta dilengkapi notifikasi otomatis. Sistem ini diharapkan meningkatkan efisiensi pemantauan dan produktivitas budidaya hidroponik.
Kata kunci: esp32; hidroponik; internet of things; kualitas air; monitoring real-time
Abstract:Abstract: Libraries function not only as information centers but also as literacy spaces that require an orderly and communicative service environment. One supporting service in fostering such an environment is the delivery…
ery of literacy greetings to visitors. In practice, greetings are commonly delivered manually or through conventional bells, leading to inconsistency and dependence on staff availability. This study was conducted at the Amir Machmud Library, Ministry of Home Affairs, Jakarta, Indonesia, aiming to design and evaluate a Python-based scheduled audio alarm system for automated literacy greetings. An applied experimental method was employed, including system design, Python script development, scheduling configuration using Windows Task Scheduler, and direct system testing on a library computer connected to ceiling speakers. The system requires initial execution via Command Prompt (CMD) when the computer is powered on, after which it operates automatically according to predefined schedules. Testing results demonstrate that the system performs scheduled audio playback accurately and operates stably without further manual intervention. The findings indicate that the proposed system provides a practical and efficient solution to enhance service consistency and support a structured and conducive literacy environment in the library.
Keywords: scheduled audio alarm; library automation; python; literacy greeting.
Abstrak: Perpustakaan tidak hanya berfungsi sebagai pusat informasi, tetapi juga sebagai ruang literasi yang memerlukan suasana layanan yang tertib dan komunikatif. Salah satu bentuk dukungan layanan tersebut adalah penyampaian sapaan literasi kepada pengunjung. Dalam praktiknya, penyampaian sapaan masih dilakukan secara manual atau menggunakan bel konvensional sehingga kurang konsisten dan bergantung pada petugas. Penelitian ini dilaksanakan di Perpustakaan Amir Machmud, Kementerian Dalam Negeri, Jakarta, Indonesia, dengan tujuan merancang dan menguji sistem alarm audio terjadwal berbasis Python sebagai media penyampaian sapaan literasi. Metode yang digunakan adalah metode eksperimental terapan melalui tahapan perancangan sistem, pengembangan skrip Python, konfigurasi penjadwalan menggunakan Windows Task Scheduler, serta pengujian langsung pada komputer perpustakaan yang terhubung dengan speaker plafon. Sistem bekerja dengan mekanisme inisialisasi awal melalui Command Prompt (CMD) saat komputer dinyalakan, kemudian selanjutnya berjalan otomatis sesuai jadwal yang telah ditentukan. Hasil pengujian menunjukkan bahwa sistem mampu memutar audio secara konsisten dan stabil pada waktu yang telah diatur tanpa intervensi lanjutan dari petugas. Dengan demikian, sistem ini dapat menjadi solusi sederhana dan efisien untuk mendukung terciptanya suasana literasi yang lebih terstruktur dan kondusif di lingkungan perpustakaan.
Kata kunci: alarm audio terjadwal; otomasi perpustakaan; python; sapaan literasi.
Abstract:Abstract: Academic achievement mapping is an important process in higher education to support effective academic monitoring and guidance. In practice, student grouping is often conducted manually by academic staff using…
simple criteria such as Grade Point Average (GPA) thresholds and subjective judgment, without systematic data analysis. This study aims to apply the Fuzzy C-Means (FCM) clustering algorithm to objectively group students based on their academic achievement levels. The dataset consists of academic records from 179 sixth-semester students of the Computer Science Study Program at Universitas Islam Negeri Sumatera Utara, where 160 eligible students are processed in the FCM calculation. Three variables are used: cumulative GPA, total completed credits, and the total number of low grades (D/E). The FCM algorithm automatically performs the mapping and groups students into three categories, namely excellent, stable, and at-risk students. Cluster quality is evaluated using the Silhouette Score and Davies–Bouldin Index, showing satisfactory clustering performance. The results indicate that the proposed approach provides a data-driven and objective basis for academic decision support.
Keywords: academic achievement; clustering; fuzzy c-means; student
Abstrak: Pemetaan pencapaian akademik mahasiswa merupakan proses penting dalam pendidikan tinggi untuk mendukung pemantauan dan pembinaan akademik yang tepat sasaran. Dalam praktiknya, pengelompokan mahasiswa masih sering dilakukan secara manual oleh pihak akademik berdasarkan kriteria sederhana, seperti batasan Indeks Prestasi Kumulatif (IPK) dan penilaian subjektif, tanpa analisis data yang sistematis. Penelitian ini bertujuan menerapkan algoritma Fuzzy C-Means (FCM) untuk mengelompokkan mahasiswa secara objektif berdasarkan tingkat pencapaian akademik. Data penelitian berasal dari 179 mahasiswa semester enam Program Studi Ilmu Komputer Universitas Islam Negeri Sumatera Utara, dengan 160 mahasiswa memenuhi kriteria dan diproses menggunakan algoritma FCM. Variabel yang digunakan meliputi IPK kumulatif, jumlah SKS yang telah ditempuh, dan total nilai rendah (D/E). Proses pemetaan sepenuhnya dilakukan oleh algoritma FCM dan menghasilkan tiga kategori mahasiswa, yaitu unggul, stabil, dan berisiko. Evaluasi menggunakan Silhouette Score dan Davies–Bouldin Index menunjukkan kualitas pengelompokan yang cukup baik.
Kata kunci: fuzzy c-means; clustering; mahasiswa; pencapaian akademik
Abstract:Abstract: The implementation of dress code regulations in university environments is generally still carried out conventionally, requiring significant time and effort and potentially leading to subjective assessments. This…
is study develops an automatic student dress code compliance detection system using computer vision based on the YOLOv8 model. The dataset consists of 1,800 annotated images divided into eight clothing categories, split into 78% training (1,404 images), 14% validation (254 images), and 8% testing (143 images). All images underwent preprocessing and data augmentation before training the YOLOv8 model with an input size of 640×640 pixels for 50 epochs. During testing, the YOLOv8 model achieved an overall performance of Precision 0.844, Recall 0.773, F1-Score 0.802, and mAP@0.5 0.841, and was able to detect clothing objects with good accuracy and stable performance under various image conditions. The system was integrated with a Flask-based backend and a web-based frontend to enable real time detection and compliance classification, with a response time of less than 2 seconds, supporting automatic and consistent identification of student dress code compliance as “Compliant” or “Violation.”
Keywords: compliance detection; computer vision; dress code regulations; real time detection; YOLOv8.
Abstrak: Penerapan aturan berpakaian di lingkungan kampus umumnya masih dilakukan secara konvensional sehingga membutuhkan waktu dan tenaga yang relatif besar serta berpotensi menimbulkan subjektivitas penilaian. Penelitian ini bertujuan mengembangkan sistem pendeteksi kepatuhan berpakaian mahasiswa secara otomatis berbasis visi komputer menggunakan model YOLOv8. Dataset yang digunakan terdiri dari 1.800 citra beranotasi yang terbagi ke dalam 8 kategori pakaian, dengan pembagian data sebesar 78% data latih (1.404 citra), 14% data validasi (254 citra) dan 8% data uji (143 citra). Seluruh citra diproses melalui tahapan pre-processing dan data augmentation, kemudian digunakan untuk melatih model YOLOv8 dengan ukuran input 640×640 piksel selama 50 epoch. Pada tahap pengujian, model mencapai performa keseluruhan dengan Precision 0.844, Recall 0.773, F1-Score 0.802, dan mAP@0.5 0.841, serta mampu mendeteksi objek pakaian dengan akurasi baik dan performa stabil pada berbagai kondisi citra. Sistem kemudian diintegrasikan dengan backend berbasis Flask dan frontend web untuk mendukung proses deteksi waktu nyata dan klasifikasi kepatuhan, dengan waktu respons sistem kurang dari 2 detik, sehingga mampu mengidentifikasi status kepatuhan berpakaian mahasiswa ke dalam kategori “Aman” dan “Melanggar Aturan” secara otomatis dan konsisten.
Kata kunci: aturan berpakaian; deteksi waktu nyata; pendeteksi kepatuhan; visi komputer; YOLOv8.
Abstract:Abstract: The rapid growth of the digital music industry requires accurate music genre classification systems to enhance user experience in streaming services. This study compares a domain-specific Long Short-Term Memory…
(LSTM) network with three Large Language Models (LLMs)—HuBERT, WavLM, and WAV2Vec 2.0—for Music Genre Classification (MGC). The LSTM model was trained using Mel-spectrograms transformed from the GTZAN dataset, while the LLMs were fine-tuned using a smaller set of raw audio samples due to computational constraints. All models were tested on datasets with identical genre labels to ensure a fair evaluation. Results show that the LSTM model achieved the highest accuracy of 97.10%, outperforming HuBERT (86.00%), WavLM (83.00%), and WAV2Vec 2.0 (80.00%). The LSTM demonstrated superior generalization and stability without overfitting, while the LLMs struggled to differentiate between genres with similar acoustic characteristics. These findings indicate that general-purpose pre-trained models, although powerful, are less effective in music-specific tasks due to domain mismatch. Therefore, incorporating music-specific features and architectures remains essential for achieving higher accuracy and reliability in automatic genre classification systems.
Keywords: audio large language models; comparative deep learning; music genre classification.
Abstrak: Pertumbuhan industri musik digital yang pesat menuntut sistem klasifikasi genre musik yang akurat untuk meningkatkan pengalaman pengguna dalam layanan streaming. Penelitian ini dilatarbelakangi oleh perkembangan pesat model pembelajaran mendalam, khususnya jaringan LSTM dan model bahasa berskala besar LLM seperti HuBERT, WavLM, dan WAV2Vec 2.0, yang telah menunjukkan kemampuan representasi audio yang kuat. Tujuan penelitian ini ini membandingkan jaringan Long Short-Term Memory (LSTM) khusus domain dengan tiga model Large Language Models (LLM)—HuBERT, WavLM, dan WAV2Vec 2.0—untuk tugas Klasifikasi Genre Musik (MGC). Metode penelitian melibatkan pelatihan LSTM menggunakan data Mel-spectrogram hasil transformasi dari dataset GTZAN, sementara LLM disesuaikan (fine-tuning) menggunakan data audio mentah dalam jumlah lebih kecil karena keterbatasan komputasi. Seluruh model diuji pada dataset dengan label genre yang sama untuk memastikan evaluasi yang adil. Hasil penelitian menunjukkan bahwa model LSTM mencapai akurasi tertinggi sebesar 97,10%, sedangkan model HuBERT, WavLM, dan WAV2Vec 2.0 masing-masing memperoleh 86,00%, 83,00%, dan 80,00%. Model LSTM menunjukkan kemampuan generalisasi yang lebih baik tanpa overfitting, sedangkan model LLM cenderung kesulitan membedakan genre dengan karakteristik akustik yang mirip. Kesimpulan penelitian ini adalah ketidaksesuaian domain secara signifikan membatasi performa model umum saat diterapkan pada tugas berbasis musik. Oleh karena itu, penggunaan fitur dan arsitektur khusus musik sangat penting dalam membangun sistem klasifikasi genre yang lebih akurat.
Kata kunci: klasifikasi genre musik; model bahasa besar; perbandingan pembelajaran mendalam.
Abstract:Abstract: The development of digital learning systems requires not only effective content delivery but also database consistency and performance, particularly when used at scale by lecturers and students. Weaknesses in database…
atabase design can lead to data duplication, relational violations, and transaction failures that compromise system reliability. This study designed the Royal Mengajar application using PHP and MySQL, supported by JavaScript, HTML, and Bootstrap 5. The Crowdsourced Academic Content model enables lecturers to contribute learning materials openly, while students evaluate them through a user rating system. The objective of this research is to design and optimize the database architecture of the Royal Mengajar application by implementing multiple control mechanisms—namely views, triggers, transactions, and constraints—to enhance data efficiency, consistency, and integrity in digital learning environments. Database optimization focuses on the use of views to improve query efficiency, triggers to maintain automatic consistency, transactions to ensure atomicity in multi-table operations, and constraints to preserve data integrity. The results show that views reduced the average query execution time to 0.12 seconds, triggers maintained consistency without manual intervention, and constraints achieved 100% referential integrity. The application of these mechanisms significantly improved system speed, reduced data redundancy, and enhanced information reliability, thus reinforcing the sustainability of Royal Mengajar as a community-driven learning platform
Keywords: crowdsourced academic content; constraint; database optimization; trigger.
Abstrak: Pengembangan sistem pembelajaran digital tidak hanya menuntut penyajian materi, tetapi juga konsistensi serta kinerja basis data ketika sistem digunakan secara masif oleh dosen dan mahasiswa. Kelemahan rancangan database dapat menimbulkan duplikasi data, pelanggaran relasi, dan kegagalan transaksi yang memengaruhi keandalan sistem. Penelitian ini merancang aplikasi Royal Mengajar berbasis PHP dan MySQL dengan dukungan JavaScript, HTML, dan Bootstrap 5. Model Crowdsourced Academic Content memungkinkan dosen berkontribusi secara terbuka, sedangkan mahasiswa melakukan evaluasi melalui user rating system. Tujuan penelitian ini adalah untuk merancang dan mengoptimalkan basis data aplikasi Royal Mengajar melalui penerapan berbagai mekanisme pengendali, seperti view, trigger, transaction, dan constraint, guna meningkatkan efisiensi, konsistensi, dan integritas data dalam sistem pembelajaran digital. Optimalisasi database difokuskan pada penerapan view untuk efisiensi query, trigger untuk menjaga konsistensi otomatis, transaction untuk memastikan atomicity pada operasi multi-tabel, serta constraint guna menjamin integritas data. Hasil pengujian menunjukkan view menurunkan rata-rata waktu eksekusi query menjadi 0,12 detik, trigger menjaga konsistensi tanpa intervensi manual, dan constraint memastikan integritas referensial tercapai 100%. Penerapan mekanisme ini berdampak pada peningkatan kecepatan sistem, berkurangnya redundansi, serta keandalan informasi yang lebih tinggi, sehingga mendukung keberlanjutan Royal Mengajar sebagai platform pembelajaran berbasis kontribusi komunitas.
Kata kunci: basis data; optimasi; trigger; constraint; crowdsourced academic content.
Abstract:Abstract: The rice plant, Oryza sativa, is a major food source in Indonesia. This plant is processed into rice, a staple food for the Indonesian people. Rice growth is crucial to ensure the rice produced is of good quality.…
ty. One part of the rice plant that is susceptible to disease is the leaves, which can inhibit growth and reduce rice quality. Therefore, early detection and accurate classification of rice diseases are crucial to minimize these negative impacts. This has driven the development of a Deep Learning model capable of high-performance automatic classification. This study aims to create a rice leaf classification model using the CNN algorithm and several transfer learning architectures such as ResNet101, VGG16, and Xception. A dataset of 859 rice leaf images collected from the Kaggle website was then processed using augmentation techniques to a total of 2,439 images, plus 215 smartphone photos for external data validation. Thus, the total dataset increased to 2,656 images, covering four categories: leafblast, brownspot, healthy, and hispa. The model was processed in two stages: on the initial dataset (Non-Augmented Dataset) and the Augmented Dataset. The best experimental results were obtained using the ResNet architecture, with a training accuracy of 96.17% and a validation accuracy of 95.22%. Based on the research results, the rice plant disease classification model using deep learning demonstrated good performance.
Keywords: convolutional neural network; deep learning; fine-tuning; image classification; resnet; rice plant
Abstract:Abstract: This study aims to develop a Deep Learning-based Traffic Flow Detector to automatically and accurately observe traffic flow. Conventional traffic observation is often conducted manually or via CCTV, but it is prone…
rone to human error and difficult to use for real-time trend analysis. In this study, the YOLOv4 method is used to detect four types of vehicles (cars, motorcycles, buses, trucks). To continuously track vehicle movement and address occlusion issues, the Deep SORT algorithm is implemented. The YOLOv4 model used is a pre-trained model and was tested on seven CCTV video recordings obtained from the official website of the Pekanbaru City Transportation Department. The system was implemented on a limited device, the Nvidia Jetson Nano, as a simulation of direct CCTV integration. Test results showed a highest precision of 98%, but the maximum accuracy achieved was only 26%. This low accuracy is influenced by several factors, including video resolution, detection model quality, and lighting conditions. Nevertheless, the system demonstrates potential to support future traffic management and engineering decisions but still requires further optimization, including improving video resolution and quality, retraining the model with a more representative local dataset, using lighter and more accurate detection models, and optimizing the tracking algorithm.
Keywords: deep learning; deepsort; NVIDIA Jetson NANO; traffic flow; YOLOv4
Abstrak: Penelitian ini bertujuan mengembangkan Traffic Flow Detector berbasis Deep Learning untuk mengobservasi arus lalu lintas secara otomatis dan akurat. Observasi lalu lintas konvensional sering dilakukan secara manual atau melalui CCTV, namun rentan terhadap human error dan sulit digunakan untuk menganalisis tren secara real-time. Pada penelitian ini digunakan metode YOLOv4 untuk mendeteksi empat jenis kendaraan (mobil, motor, bus, truk). Untuk melacak pergerakan kendaraan secara berkelanjutan dan mengatasi masalah occlusion, digunakan algoritma Deep SORT. Model YOLOv4 yang digunakan merupakan pre-trained model dan diujikan pada tujuh rekaman video CCTV yang diambil dari situs resmi Dinas Perhubungan Kota Pekanbaru. Sistem ini diimplementasikan pada perangkat terbatas Nvidia Jetson Nano sebagai simulasi penerapan langsung pada CCTV. Hasil pengujian menunjukkan presisi tertinggi mencapai 98%, namun akurasi tertingginya hanya sebesar 26%. Rendahnya akurasi dipengaruhi oleh beberapa faktor seperti resolusi video, kualitas model deteksi, serta kondisi pencahayaan. Meski demikian, sistem ini menunjukkan potensi untuk membantu pengambilan keputusan dalam manajemen dan rekayasa lalu lintas di masa depan, namun masih membutuhkan optimasi lebih lanjut, seperti peningkatan kualitas video input, pelatihan ulang model dengan dataset lokal, penggunaan model deteksi yang lebih ringan dan akurat serta pengoptimalan algoritma pelacakan.
Kata kunci: deep learning deepsort; Nvidia Jetson Nano; traffic flow; YOLOv4