Garuda - Garba Rujukan Digital

Pengenalan Pola Karakter Bahasa Jepang Hiragana Menggunakan 2D Convolutional Neural Network Mellysa Margarita Susilo; Daniel Martomanggolo Wonohadidjojo; Nehemia Sugianto
Jurnal Informatika dan Sistem Informasi Vol. 3 No. 2 (2017): Jurnal Informatika dan Sistem Informasi
Publisher : Universitas Ciputra Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (1048.914 KB)

Komik disukai oleh banyak orang di seluruh dunia. Komik Jepang atau yang biasa disebut manga dapat ditemukan di internet, tapi tidak semua orang dapat membacanya karena berbahasa Jepang. Terdapat tim non-profit yang bersedia menerjemahkan manga namun karena waktu dan tenaga yang terbatas maka tidak semua manga dapat diterjemahkan. Oleh sebab itu, walaupun terdapat banyak manga di internet, tidak semua orang dapat membacanya. Sehingga menimbulkan kekecewaan pada pembaca yang menantikannya. Masalah ini dapat diselesaikan dengan menggunakan pattern recognition dan algoritma convolutional neural network. Pada penelitian ini akan menggunakan media gambar komik. Langkah pertama yang dilakukan adalah mengambil tulisan Jepang dari balon teks pada manga. Setelah tulisan Jepang didapatkan, dilakukan feature extraction untuk dianalisa dan disimpan polanya untuk dibandingkan dengan bank data tulisan yang telah dipersiapkan sehingga dapat ditentukan karakter yang terdapat pada sumber gambar dikenali sebagai apa. Untuk melatih agar pattern recognition menghasilkan hasil yang maksimal, dikembangkanlah dengan algoritma convolutional neural network agar aplikasi dapat berjalan secara mandiri dan semakin pintar dalam mengenali pola. Akan terdapat dua aplikasi, aplikasi pertama adalah Element Extractor from Manga (Japanese Comic) yang dikembangkan dengan bahasa pemograman C# dan library AForge. Aplikasi kedua adalah program pattern recognition yang dikembangkan dengan MATLAB. Tes dilakukan dengan 10 input yang berbeda untuk tiap tahapnya. Element Extractor from Manga berhasil mengekstrak 88% frame komik, 91% balon teks, dan 46% karakter. Pelatihan convolutional neural network mencapai akurasi 96.2% dan tes cross validation mencapai akurasi 86%.

The Effects of Preprocessing Techniques on Nasnetmobile's Performance for Classifying Knee Osteoarthritis Based on the Kellgren-Lawrence System Wiradinata, Marcell Jeremy; Wonohadidjojo, Daniel Martomanggolo
Journal of Applied Informatics and Computing Vol. 8 No. 2 (2024): December 2024
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v8i2.8713

Knee osteoarthritis (KOA) is a degenerative joint disorder characterized by the progressive deterioration of protective cartilage at the ends of bones, leading to pain and limited mobility. Deep learning provides an effective approach to classify whether X-ray images indicate the presence of KOA; however, dataset preprocessing techniques can enhance the efficacy of deep learning models. This study highlights the importance of preprocessing techniques in improving image contrast, particularly in utilizing the NASNetMobile model to assess the severity of KOA through X-ray images. KOA classification based on the Kellgren-Lawrence system consists of five severity levels; however, simplifying it into two categories can improve the performance of deep learning models. By fine-tuning parts of the NASNetMobile model and using the Nadam optimizer, the model initially achieved only 59% validation accuracy. However, by applying various preprocessing techniques, the model's validation accuracy improved to 80%.

Penerapan Algoritma Yolov4-Tiny Dan Efficientnetv2-S Untuk Deteksi Kesegaran Ikan Gurami Natadjaja, Hans Richard Alim; Wonohadidjojo, Daniel Martomanggolo
Jurasik (Jurnal Riset Sistem Informasi dan Teknik Informatika) Vol 8, No 2 (2023): Edisi Agustus
Publisher : STIKOM Tunas Bangsa Pematangsiantar

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30645/jurasik.v8i2.633

Fish production is one of the largest in Indonesia. Among many fishery products, Gurami fish is one of the most widely processed by society. However, the decrease in the freshness of fishery products is susceptible to occur when they reach the hands of consumers. Several methods for detecting fish freshness have been applied to assist in obtaining fresh fish products. However, some of these methods have a lack of accuracy rate, not all of them were developed to detect gurami freshness. Therefore, a solution is needed to help people detect the freshness of Gurami fish and make it accessible on mobile devices. This solution can be realized by using Deep Learning methods. The methods proposed in this study use the Convolutional Neural Network (CNN) by utilizing the YOLOv4-Tiny algorithm to detect the Region of Interest (ROI) and the EfficientNetV2-S architecture freshness classification on Gurami fish images. The training process using both methods can produce an ROI detection model with a mean average precision of 93,58% and a classification model with an accuracy rate of 93%

Net Zero Healthy Building: Defining the Terms and Criteria with Bibliometric Analysis Susan, Susan; Wardhani, Dyah Kusuma; Ariyanto, Yusuf; Wonohadidjojo, Daniel Martomanggolo; Harianto, Eric
EMARA: Indonesian Journal of Architecture Vol. 9 No. 1 (2023): August 2024 ~ October 2024
Publisher : Universitas Islam Negeri Sunan Ampel Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29080/eija.v9i1.1419

Net Zero Energy Buildings (NZeBs) are a key design and construction paradigm due to sustainable building practices. NZeBs emphasize occupant well-being and reducing dependence on non-renewable energy sources in building design. A rigorous bibliometric investigation will define and standardize NZeBs in this work. This study uses data mining, VOSviewer bibliometric analysis, and a comprehensive literature search to identify key themes and criteria for understanding Net Zero Energy Buildings. The findings show that building system boundaries, passive design, energy efficiency assessments, and adaptive thermal comfort principles help conceptualize Net Zero Energy Buildings. This research improves our understanding of NZeBs and lays the groundwork for future research on sustainable building practices and healthier interior environments by merging data from other sectors.

Pengenalan Pola Karakter Bahasa Jepang Hiragana Menggunakan 2D Convolutional Neural Network Susilo, Mellysa Margarita; Wonohadidjojo, Daniel Martomanggolo; Sugianto, Nehemia
Jurnal Informatika dan Sistem Informasi Vol. 7 No. 2 (2021): Jurnal Informatika dan Sistem Informasi
Publisher : Universitas Ciputra Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar

Komik disukai oleh banyak orang di seluruh dunia. Komik Jepang atau yang biasa disebut manga dapat ditemukan di internet, tapi tidak semua orang dapat membacanya karena berbahasa Jepang. Terdapat tim non-profit yang bersedia menerjemahkan manga namun karena waktu dan tenaga yang terbatas maka tidak semua manga dapat diterjemahkan. Oleh sebab itu, walaupun terdapat banyak manga di internet, tidak semua orang dapat membacanya. Sehingga menimbulkan kekecewaan pada pembaca yang menantikannya. Masalah ini dapat diselesaikan dengan menggunakan pattern recognition dan algoritma convolutional neural network. Pada penelitian ini akan menggunakan media gambar komik. Langkah pertama yang dilakukan adalah mengambil tulisan Jepang dari balon teks pada manga. Setelah tulisan Jepang didapatkan, dilakukan feature extraction untuk dianalisa dan disimpan polanya untuk dibandingkan dengan bank data tulisan yang telah dipersiapkan sehingga dapat ditentukan karakter yang terdapat pada sumber gambar dikenali sebagai apa. Untuk melatih agar pattern recognition menghasilkan hasil yang maksimal, dikembangkanlah dengan algoritma convolutional neural network agar aplikasi dapat berjalan secara mandiri dan semakin pintar dalam mengenali pola. Akan terdapat dua aplikasi, aplikasi pertama adalah Element Extractor from Manga (Japanese Comic) yang dikembangkan dengan bahasa pemograman C# dan library AForge. Aplikasi kedua adalah program pattern recognition yang dikembangkan dengan MATLAB. Tes dilakukan dengan 10 input yang berbeda untuk tiap tahapnya. Element Extractor from Manga berhasil mengekstrak 88% frame komik, 91% balon teks, dan 46% karakter. Pelatihan convolutional neural network mencapai akurasi 96.2% dan tes cross validation mencapai akurasi 86%.

CLASSIFICATION OF BONE FRACTURES IN THE WRIST AND HAND USING DENSENET AND XCEPTION Nusantara, Michelle Swastika Bianglala; Wonohadidjojo, Daniel Martomanggolo
JIKO (Jurnal Informatika dan Komputer) Vol 8, No 1 (2025)
Publisher : Universitas Khairun

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33387/jiko.v8i1.9201

This study aims to apply Convolutional Neural Network (CNN) using DenseNet and Xception to classify fracture in the wrist and hand bones, while utilizing transfer learning to enhance model's performance. Accurate diagnosis and successful treatment of bone fractures depend on early identification, which lowers the likelihood of long-term issues such avascular necrosis or non-union. The research utilized data from two publicly available musculoskeletal radiography datasets and employed deep learning techniques with the Keras framework. DenseNet was selected for wrist image analysis due to its dense connectivity, which preserves information from previous layers, while Xception was chosen for hand bone image analysis because of its ability to identify complex patterns using depthwise separable convolutions. Transfer learning was implemented to accelerate training and improve accuracy. The DenseNet model achieved a test accuracy of 97.5% for wrist classification, while the Xception model reached 92% accuracy for hand bone classification. By tailoring CNN architectures to specific radiographic images and employing transfer learning, this study demonstrates significant potential for improving diagnostic precision in clinical situations. Furthermore, the findings can support medical personnel in detecting bone fractures more efficiently and accurately, ultimately expediting clinical decision-making and improving patient care.

Classification of Skin Diseases Using Transfer Learning with ResNet-50 Architecture and Data Preprocessing Using Real-ESRGAN and Wiener Filter Jiemesha, Micheila; Wonohadidjojo, Daniel Martomanggolo
Intelligent System and Computation Vol 7 No 1 (2025): INSYST: Journal of Intelligent System and Computation
Publisher : Institut Sains dan Teknologi Terpadu Surabaya (d/h Sekolah Tinggi Teknik Surabaya)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52985/insyst.v7i1.399

The skin is a vital organ which serves as a barrier against external factors, yet it’s highly susceptible to diseases. These diseases are often presented as lesions with similar appearances, making it difficult to be diagnosed and prone to human errors. To address this challenge, this study uses Deep Learning, particularly the ResNet-50 architecture using Transfer Learning, to classify skin diseases from lesion. In this study, data augmentation is implemented to increase dataset size, thus improving model performance and preventing overfitting. Data is then preprocessed using Real-ESRGAN to enhance resolution and the Wiener Filter to sharpen the features. Adam optimizer is used to further enhance the model’s performance. Hyperparameter tuning is also implemented to optimize the model parameters, and dropout regularization is applied to enhance the model's ability to be able to accurately classify unseen data. The model managed to achieve a high accuracy of 99.09%, with 0.96 precision, 0.95 recall, and 0.95 F1-score. These results demonstrate the effectiveness of combining Real-ESRGAN and Wiener Filter with the ResNet-50 architecture and the Adam optimizer in developing a robust model for skin disease classification. This approach offers a promising tool for healthcare professionals which may help reduce human error in dermatological diagnosis.

Evaluasi Performa Panel Surya Terintegrasi Bangunan berdasarkan Standar Greenship: Menuju Bangunan Sekolah Net Zero Energi Susan, Susan; Wardhani, Dyah Kusuma; Ariyanto, Yusuf; Wonohadidjojo, Daniel Martomanggolo; Harianto, Eric
EMARA: Indonesian Journal of Architecture Vol. 8 No. 1 (2022): Vol. 8 No. 1 (2022): EIJA August-October edition
Publisher : Universitas Islam Negeri Sunan Ampel Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29080/eija.v8i1.1442

generated from a variety of sources, both renewable and nonrenewable. Switching from nonrenewable to renewable energy sources is one of many strategies that can be used to achieve net-zero buildings. In Indonesia, this strategy is very feasible due to its abundant renewable energy resources, particularly solar energy. This research presents a school building as the proposed case. The school, SCK Citra Garden, is chosen as the pilot project due to its access to solar radiation and its minimum shading conditions. Using Helioscope software, BIPV modelling was simulated on its roof, and the electrical energy output from BIPV was calculated. The substitution percentages of BIPV energy output for conventional electrical energy consumed by the building were then measured. This percentage was compared to the National Energy Mix target and Greenhouse Gas Standard to assess its performance towards net-zero school buildings. The result shows that BIPV has a good performance. Even though the substitution percentage is still below the national energy mix target, it exceeds the greenhouse gas standard target for on-site renewable energy tools.

Sistem Deteksi dan Klasifikasi Truk Air Menggunakan YOLO v5 dan EfficientNet-B4 Kurniawan, Ardian; Wonohadidjojo, Daniel Martomanggolo
Intelligent System and Computation Vol 5 No 2 (2023): INSYST: Journal of Intelligent System and Computation
Publisher : Institut Sains dan Teknologi Terpadu Surabaya (d/h Sekolah Tinggi Teknik Surabaya)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52985/insyst.v5i2.356

Kegiatan pencatatan dalam usaha pengisian air yang dilakukan dengan menggunakan truk air mengalami masalah karena kesalahan manusia (human error) misalnya pencatatan yang terlewat dan efisiensi waktu yang diperlukan. Untuk itu diperlukan otomatisasi sistem dengan menggunakan teknologi. Untuk mengatasi masalah tersebut, pada penelitian ini digunakan metode yang termasuk dalam Computer Vision dengan penggunaan algoritma Object Detection dan Classification. Pada penelitian ini dibangun suatu sistem yang mengambil frame video menggunakan CCTV yang kemudian dimasukkan pada algoritma object detection dengan arsitektur YOLOv5 (You Only Look Once versi 5). Hasil deteksi kemudian diklasifikasikan dengan menggunakan algoritma dengan arsitektur EfficientNet-B4. Hasil klasifikasi tersebut akan menentukan secara spesifik truk air yang mana yang sedang melakukan pengisian dan dicatat. Kemudian rekapitulasi hasil pencatatan tersebut dikirimkan dengan menggunakan aplikasi messaging Telegram menggunakan library Tkinter kepada pemilik usaha yang mengambil air tersebut. Rekapitulasi tersebut kemudian digunakan oleh sang pemilik usaha dalam memantau usaha tersebut dan melakukan pembayaran sesuai dengan jumlah pengambilan air. Hasil pengujian untuk object detection dan classification dengan menggunakan evaluation metrics menunjukkan bahwa metode tersebut berhasil melakukan deteksi dan klasifikasi dengan baik. Pengujian keseluruhan sistem menunjukkan bahwa semua tahap pengujian berhasil dilakukan dengan baik. Hal ini menunjukkan bahwa sistem tersebut dapat digunakan untuk mengatasi masalah yang dihadapi.

Implementation of DenseNet Architecture With Transfer Learning to Classify Mango Leaf Diseases Likorawung, Marsha Alexis; Wonohadidjojo, Daniel Martomanggolo
Intelligent System and Computation Vol 6 No 2 (2024): INSYST: Journal of Intelligent System and Computation
Publisher : Institut Sains dan Teknologi Terpadu Surabaya (d/h Sekolah Tinggi Teknik Surabaya)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52985/insyst.v6i2.401

Mango plants (Mangifera indica) are a significant export commodity in the horticultural industry, offering numerous nutritional and economic benefits. They are rich in essential micronutrients, vitamins, and phytochemicals, contributing to their high demand globally. However, mango plants are susceptible to various diseases that can severely impact their yield and quality. These diseases pose a challenge to mango farmers, many of whom struggle to identify and treat them effectively, leading to potential harvest failures. This study aims to address this challenge by implementing a Deep Learning approach to classify diseases in mango leaves. Specifically, the research utilizes a Convolutional Neural Network (CNN) with DenseNet architecture, known for its efficiency in image classification tasks. The study incorporates Contrast Limited Adaptive Histogram Equalization (CLAHE) for image preprocessing to enhance detail and improve the model’s performance. Transfer Learning is utilized to optimize the DenseNet model, leveraging a pre-trained model to achieve high accuracy even with a relatively small dataset. The dataset used in this research comprises 4000 labeled images of mango leaves, covering seven disease categories and healthy leaves. These images include common diseases such as Anthracnose, Dieback, Powdery Mildew, Red Rust, Cutting Weevil, Bacterial Canker, and Sooty Mould. The DenseNet model achieved an overall accuracy of 99.5% in classifying mango leaf diseases.

Title

Found 24 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 24 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 24 Documents
Search