Article Per Year (5 Year)
p-Index From 2021 - 2026
1.513
P-Index
This Author published in this journals
All Journal Semantik Jurnal Pseudocode JAIS (Journal of Applied Intelligent System) Prosiding Seminar Nasional IPTEK Terapan (SENIT) 2016 Pengembangan Sumber Daya Lokal Berbasis IPTEK Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) RABIT: Jurnal Teknologi dan Sistem Informasi Univrab Abdimasku : Jurnal Pengabdian Masyarakat Jurnal Pengabdian kepada Masyarakat Nusantara Jurnal Nasional Teknik Elektro dan Teknologi Informasi Scientific Journal of Informatics
Catur Supriyanto
Universitas Dian Nuswantoro
Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Economics, Econometrics & Finance Electrical & Electronics Engineering Energy Engineering Health Professions Industrial & Manufacturing Engineering Social Sciences Other

Published : 13 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 13 Documents
Search

 1   2 
Hybrid DenseNet201-ViT Architecture for Robust Solid Waste Classification Lailia Rahmawati; Catur Supriyanto
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 10 No 5 (2026): October 2026
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v10i5.7669

Abstract

Automatic sorting of solid waste is difficult because discarded items are often badly deformed and some material classes, such as transparent plastic and glass, look almost identical. Conventional Convolutional Neural Networks (CNNs) mainly learn local patterns; as a result, they tend to overlook global context and to overfit when only a small and unevenly distributed dataset is available. This study develops a hybrid deep learning model that pairs the dense local feature learning of DenseNet201 with the global context modeling of a Vision Transformer (ViT) in order to make waste classification more robust. An ImageNet-pretrained DenseNet201 serves as the backbone, and its high-level feature maps are converted into a sequence of patch tokens that is processed by a custom ViT encoder. Experiments were conducted on a public garbage image dataset with six unevenly represented classes, namely cardboard, glass, metal, paper, plastic, and trash. On-the-fly data augmentation and early stopping were employed during training to limit overfitting. The hybrid model reached an overall accuracy of 87% and a macro F1-score of 0.85, exceeding the CNN baselines trained under identical settings. It was especially accurate on fibrous materials, with precision values of 0.95 for cardboard and 0.89 for paper. These results indicate that coupling self-attention with densely connected convolutional layers alleviates the main weaknesses of pure CNNs and provides an accurate, interpretable, and computationally efficient basis for practical automated waste sorting systems.
Comparative Analysis of YOLOv5 and YOLOv8 Cigarette Detection in Social Media Content Cinantya Paramita; Catur Supriyanto; Amalia; Khalivio Rahmyanto Putra
Scientific Journal of Informatics Vol. 11 No. 2: Regular Issue
Publisher : Universitas Negeri Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15294/sji.v11i2.2808

Abstract

Purpose: Addresses the pressing public health concern of tobacco product portrayal on social media, which significantly influences the younger demographic by glamorizing smoking culture. The purpose is to compare the capabilities of YOLOv5 and YOLOv8 models in detecting and censoring cigarette-related imagery on social media platforms, aiming to reduce exposure among children and teenagers. Methods: Employing a dataset of 2,188 images collected from Twitter, this research undertook a comprehensive methodology involving data preprocessing, YOLOv5 and YOLOv8 model training, and rigorous evaluation. The study utilized mean Average Precision (mAP) and F1-Score metrics to evaluate the performance of YOLOv5 and YOLOv8 models, focusing on their precision, recall, and efficacy in detecting cigarette and cigarette pack objects. Result: The analysis highlighted YOLOv8's superiority, with a marginally higher mAP value of 0.933 compared to YOLOv5's 0.919, alongside enhanced precision and recall rates. This result underscores YOLOv8's advanced object detection capabilities, owing to its architectural innovations and anchor-free detection system. Additionally, the study confirmed the absence of significant overfitting or underfitting issues, indicating robust learning processes of the models. Novelty: Innovates in digital public health by using YOLOv5 and YOLOv8 models to automatically censor tobacco-related content on social media, effectively reducing youth exposure to such imagery. YOLOv8, in particular, exhibits marginally superior detection capabilities. The evaluation results surpass those of previous research on cigarette and cigarette burning detection, underscoring the study's significant contribution to future research and public health initiatives.
Optimizing Deep Learning Models with Custom ReLU for Breast Cancer Histopathology Image Classification Wahyu Adi Nugroho; Catur Supriyanto; Pujiono Pujiono; Guruh Fajar Shidik
Scientific Journal of Informatics Vol. 11 No. 3: Special Issue
Publisher : Universitas Negeri Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15294/sji.v11i3.12722

Abstract

Purpose: The prompt identification of breast cancer is crucial in preventing the considerable damage inflicted by this dangerous form of cancer, which is widely happened across the globe. This study seeks to refine the efficacy of a deep learning-driven approach for the precise diagnosis of breast cancer by employing diverse bespoke Rectified Linear Units (ReLU) to improve the model's performance and reduce inaccuracies within the system. Method: This study focuses on analyzing a deep learning approach utilizing the BreakHis dataset with 7,909 images, incorporating changes to the ReLU activation function across different pre-trained CNN models. It then evaluates performance through measurement such as accuracy, precision, recall, and F1-Score. Result: Based on our experiment results, it can be shown that the DenseNet201 models with a custom LeakyReLU excel beyond the typical ReLU, achieving the highest accuracy, recall, and F1-Score at 99.21%, 99.21%, and 99.11%, respectively. Simultaneously, ResNet152, utilizing LessNegativeReLU (α=0.05), achieved the highest precision at 99.11%. The VGG11 model exhibited the most notable performance enhancement, with improvements ranging from 1.39% to 1.59%. Novelty: The research is original in optimizing a model for accurate breast cancer diagnosis. The proposed model is superior to the model utilizing the default activation function. This finding indicates that the study significantly enhances performance while effectively minimizing errors, thereby necessitating further exploration into the effectiveness of the customized activation function when applied to other medical imaging modalities.
Co-Authors Abu Salam Amalia Ardytha Luthfiarta Cinantya Paramita Defri Kurniawan Dhita Aulia Octaviani Erlin Dolphina Erwin Yudi Hidayat Etika Kartikadarma Fajrian Nur Adnan Fauzi Adi Rafrastara Guruh Fajar Shidik Ika Novita Dewi Junta Zeniarja Khalivio Rahmyanto Putra Lailia Rahmawati Moch Arief Soeleman MUHAMMAD RIZKY ERI EKO JULIANTO Nur Aeni Widiastuti Oman Somantri Pujiono Pujiono Pulung Nurtantio Andono Ricardus Anggi Pramunendar Shelomita Fitriyani Sindhu Rakasiwi Stefanus Santosa Sulasmoro, Arfan Haqiqi Syahrul Idhom Utomo, Danang Wahyu Vincent Suhartono Wahyu Adi Nugroho Winarsih, Nurul Anisa Sri Yani Parti Astuti