Journal of Computer Science and Technology (JCS-TECH) https://journal.unwidha.ac.id/index.php/jcstech <p>Journal of Computer Science and Technology (JCS-TECH), published by LPPM Universitas Widya Dharma, is a peer-reviewed scientific journal dedicated to publishing original research, scholarly contributions, and innovative applications in computer science and technology. It serves as an academic platform for researchers, academics, and practitioners to disseminate research findings, technological innovations, and recent developments in these fields. JCS-TECH holds Print ISSN: 2809-1140 and Electronic ISSN: 2808-9677, is published biannually, and is accredited SINTA 5 by the Indonesian Ministry of Higher Education, Science, and Technology.</p> en-US jcstech@unwidha.ac.id (Agustinus Suradi) jcstech@unwidha.ac.id (LPPM Universitas Widya Dharma) Mon, 16 Nov 2026 09:25:44 +0700 OJS 3.3.0.13 http://blogs.law.harvard.edu/tech/rss 60 A COMPARATIVE ANALYSIS OF THE MOBILENETV2 AND RESNET50 MODELS FOR LIGHT VEHICLE IMAGE CLASSIFICATION https://journal.unwidha.ac.id/index.php/jcstech/article/view/626 <p><em>Advances in transportation technology have driven the need for vehicle identification systems capable of operating automatically, quickly, and accurately, particularly to support the implementation of automated parking systems and smart transportation systems. This study aims to compare the performance of Transfer Learning-based Convolutional Neural Network (CNN) architectures namely MobileNetV2 and ResNet50 in classifying images of cars and motorcycles. The dataset used was sourced from Kaggle and consists of a total of 1,590 images, comprising 790 car images and 800 motorcycle images. The dataset was divided into training data (80%), validation data (10%), and test data (10%). The preprocessing stage included resizing the images to 224 × 224 pixels, normalization, and data augmentation. Both models were trained for 20 epochs using the Adam optimizer with a Transfer Learning approach. Model evaluation was performed using the metrics accuracy, precision, recall, F1-score, and confusion matrix. The test results showed that the MobileNetV2 model achieved an accuracy of 98.74%, a precision of 0.99, a recall of 0.99, and an F1-score of 0.99. Meanwhile, the ResNet50 model achieved an accuracy of 79.87%, a precision of 0.80, a recall of 0.80, and an F1-score of 0.80. These results indicate that MobileNetV2 outperforms ResNet50 in classifying vehicle images. Thus, MobileNetV2 has been shown to deliver better performance and has the potential to be applied as a vehicle classification model in automated parking systems and various AI-based smart transportation applications.</em></p> Syahira Nasywa Andina Humaedi, Adinda Rahmawati, Elsha Nurdwi Baharsyah, Hanif Zaidan Sinaga Copyright (c) 2026 Journal of Computer Science and Technology (JCS-TECH) https://creativecommons.org/licenses/by-nc/4.0 https://journal.unwidha.ac.id/index.php/jcstech/article/view/626 Mon, 16 Nov 2026 00:00:00 +0700 DESIGN AND DEVELOPMENT OF A LABIRIN GAME TO ENHANCE USERS’ MEMORY BASED ON AUGMENTED REALITY (AR) https://journal.unwidha.ac.id/index.php/jcstech/article/view/426 <p><em>Technological advancements, especially in the field of digital gaming, have changed the way we interact with various aspects of life, including in the development of cognitive skills. Digital games, especially video games, now serve not only as entertainment, but also as an effective tool to improve cognitive abilities such as memory. One type of game that is beneficial in training memory is Augmented Reality-based maze games, which require players to solve problems, make strategic decisions, and maintain focus when facing obstacles. The use of Augmented Reality (AR) technology alone can take maze games to a new level, allowing for more complex, engaging, and educational maze designs as AR combines virtual objects into real environments to create dynamic and challenging interactive experiences. At SDN Sumber Kepuh 5, problems were found related to the lack of students' memory in&nbsp;in learning, especially in grade 2, which is caused by learning methods that do not suit their learning style. To address this problem, Augmented Reality (AR)-based maze games are proposed as an alternative solution that can improve students' memory and cognitive skills. Through a comparison of pretest and posttest results, it was found that the use of this game was able to improve students' memory skills, shown by a decrease in game completion time after the training session. AR maze games prove to be an interactive and fun learning medium in supporting children's cognitive development.</em></p> Yudo Bismo Utomo, Sindi April Lia, Harso Kurniadi Copyright (c) 2026 Journal of Computer Science and Technology (JCS-TECH) https://creativecommons.org/licenses/by-nc/4.0 https://journal.unwidha.ac.id/index.php/jcstech/article/view/426 Mon, 16 Nov 2026 00:00:00 +0700 PETA POSISI KEUNGGULAN KOMPETITIF BPR SYARIAH BERBASIS KEUNGGULAN TEKNOLOGI INFORMASI DAN KOMPUTER DI KABUPATEN KLATEN: SUATU KONSEPTUAL MODEL https://journal.unwidha.ac.id/index.php/jcstech/article/view/531 <p>This study aims at testing the positioning of ICT-based competitive excellences of two Sharia BPRs in <em>Klaten Regency. Through research activities, strategic marketing efforts for BPRS can be formulated. This activity culminates in the increasing of the competitiveness of Sharia BPRs in Klaten Regency towards the era of sharia industry 4.0. A quantitative Multidimensional Scaling test is conducted in this research activity. This research proposal has a novelty, namely the MDS model test which is applied directly to the observation object of two Sharia BPRs in Klaten Regency. This research seeks to create a policy brief for the best sharia banking services towards the competitive excellences of the sharia economy. The population in the study is all management, BPRS managers and the community/BPRS customers. The sampling technique for respondents uses purposive sampling of 100 respondents in two Islamic BPRs: BPRS Dharma Kuwera Klaten and BPRS Al Mabrur Klaten. Four questionnaire variable attributes are studied, including business operating environment variables, IT infrastructure development variables, IT-related management-employee knowledge variables, and regional competitiveness variables. Quantitative tests are used in the activity plan, including testing the validity and reliability of instrument items and testing the MDS model. The final stage of the research: a position map of the two Islamic BPRs is formed based on ICT competitive excellences in both qualitative and quantitative aspects. This position map of the competitive excellence model concept can be used to create competitive strategies for each Islamic BPR, thus realizing a robust Islamic banking industry model and contributing optimally to micro and small businesses in the Klaten Regency area</em></p> Beta Yudha Mahindarta, Arif Julianto Sri Nugroho, Abdul Haris, Dandang Setyawanti, Tasari, Agus Santoso Copyright (c) 2026 Journal of Computer Science and Technology (JCS-TECH) https://creativecommons.org/licenses/by-nc/4.0 https://journal.unwidha.ac.id/index.php/jcstech/article/view/531 Mon, 16 Nov 2026 00:00:00 +0700 PENGEMBANGAN SISTEM AI PARAFRASE BAHASA INFORMAL KE FORMAL MENGGUNAKAN SENTENCE-BERT SIMILARITY DAN GEMINI API https://journal.unwidha.ac.id/index.php/jcstech/article/view/623 <p><em>The use of informal language in digital communication often poses difficulties when users need to communicate formally, particularly in academic and professional contexts. This research aims to develop a system for paraphrasing informal Indonesian into formal Indonesian using the Retrieval-Augmented Style transfer (RAST) approach, which combines Sentence-BERT Similarity and the Gemini API. Sentence-BERT Similarity is used to search for example sentences with similar meanings from the dataset, whilst the Gemini API is used to generate formal paraphrases through a style transfer process. The dataset used consists of 1,110 pairs of informal and formal sentences divided into four communication contexts: lecturers, university administration, LinkedIn, and job application emails. The implementation results show that the system is capable of generating formal sentences that are more polite, clear, and appropriate to the communication context selected by the user. Manual evaluation results indicate that the system achieved an overall average score of 4.54, categorised as ‘very good’. The combination of Sentence-BERT Similarity and the Gemini API is effective in facilitating the transformation of informal language into formal language whilst preserving the core meaning of the message. This research is expected to contribute to the development of Indonesian Natural Language Processing applications, particularly in the fields of semantic Similarity and style transfer.</em></p> Maulana Farid Fatin, Muhamad Rian Hermawan, Anisa Nur Ramadhani, Hanif Zaidan Sinaga Copyright (c) 2026 Journal of Computer Science and Technology (JCS-TECH) https://creativecommons.org/licenses/by-nc/4.0 https://journal.unwidha.ac.id/index.php/jcstech/article/view/623 Mon, 16 Nov 2026 00:00:00 +0700 PERANCANGAN E-VOTING BERBASIS WEBSITE GUNA EFISIENSI PEMILIHAN ORGANISASI MAHASISWA STMIK AMIKOM SURAKARTA https://journal.unwidha.ac.id/index.php/jcstech/article/view/591 <p>Organisasi mahasiswa di STMIK Amikom Surakarta masih menggunakan metode pemilihan ketua secara konvensional yang membutuhkan kehadiran fisik. Metode ini kurang efisien akibat padatnya jadwal mahasiswa. Penelitian ini bertujuan untuk mengembangkan sistem e-voting berbasis website guna meningkatkan efisiensi dan fleksibilitas dalam pemilihan ketua organisasi mahasiswa (Ormawa). Sistem ini memungkinkan mahasiswa memberikan suara secara daring (online), sehingga meningkatkan partisipasi dan mengurangi biaya serta kebutuhan sumber daya. Selain itu, sistem ini mendukung keberlanjutan lingkungan dengan mengurangi penggunaan kertas serta memudahkan akses bagi mahasiswa dengan jadwal yang beragam. Metode pengembangan perangkat lunak menggunakan metode waterfall, yang meliputi tahapan analisis, desain sistem, pengkodean, dan pengujian. Hasil penelitian menunjukkan bahwa sistem e-voting berbasis website ini efektif meningkatkan kualitas, efisiensi, dan transparansi dalam pemilihan ketua Ormawa di STMIK Amikom Surakarta</p> Anggi Susanti, Muhammad Reno Afrido Arwinto, Yasir Galuh Adi Saputro Copyright (c) 2026 Journal of Computer Science and Technology (JCS-TECH) https://creativecommons.org/licenses/by-nc/4.0 https://journal.unwidha.ac.id/index.php/jcstech/article/view/591 Mon, 16 Nov 2026 00:00:00 +0700