Claim Missing Document
Check
Articles

Deteksi Logo Kendaraan dengan MSER-Vertical Sobel Gamma Kosala; Agus Harjoko; Sri Hartati
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 7 No 5 (2023): October 2023
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v7i5.5034

Abstract

Detecting a vehicle logo is the first step before realizing the identity of the logo. However, the detection of logos can pose difficulties due to various factors, including logo variations, differing scales and orientations, background interference, varying lighting conditions, and partial obstruction. This paper presents a vehicle logo detection method using hand-crafted features. We used a combination of Maximally Stable Extremal Region (MSER) and Vertical Sobel. We combine vertical Sobel with MSER to overcome MSER's limitation in recognizing objects of different sizes. These two features are merged using a closing morphology operation to form blobs selected as logo candidate areas. Moreover, a Support Vector Machine (SVM) is implemented to choose a logo area by analyzing each candidate's Histogram of Oriented Gradient (HOG). The proposed method was compared with other methods by implementing them on the same dataset. The significant advantage of using MSER-Vertical Sobel is its fast computation time. It is faster than other approaches that use non-handcrafted features. The test results show that the MSER-Vertical Sobel can achieve high accuracy and the fastest computation time.
Peningkatan Performa Klasifikasi Sel Darah Merah pada Pasien Talasemia Minor Latifah, Husnul; Tyas, Dyah Aruming; Harjoko, Agus
IJEIS (Indonesian Journal of Electronics and Instrumentation Systems) Vol 14, No 2 (2024): October
Publisher : IndoCEISS in colaboration with Universitas Gadjah Mada, Indonesia.

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22146/ijeis.93025

Abstract

Talasemia merupakan kelainan darah turunan yang menyebabkan rusaknya rantai hemoglobin pada eritrosit penderita. Pada kasus talasemia minor, pasien hanya menjadi pembawa gen talasemia dan tidak bergejala. Hal ini menyebabkan sedikitnya penderita talasemia minor yang terdeteksi. Saat ini, ahli hematologi harus menghitung eritrosit abnormal secara manual berdasarkan  bentuk, warna, dan tekstrur sel. Untuk itu, banyak penelitian yang memanfaatkan citra eritrosit untuk melakukan mengklasifikasi keabnormalan pada citra eritrosit secara otomatis. Namun, jumlah data yang terbatas menyebabkan salah satu jenis keabnormalan yaitu sel pensil belum terklasifikasi pada penelitian sebelumnya. Tujuan dari penelitian ini adalah mengatasi kendala tersebut dengan melakukan klasifikasi secara bertingkat dimana pada tahapan klasifikasi pertama sel pensil dikelompokkan kepada sel yang mirip, yaitu sel eliptosis terlebih dahulu. Penelitian ini menggunakan klasifier Convolutional Neural Network (CNN) pada proses klasifikasi pertama dan Support Vector Machine (SVM) pada proses klasifikasi kedua. Hasil eksperimen menujukkan klasifier CNN dengan arsitektur MiniVGGNet dan berhasil mengkasifikasi citra eritrosit ke dalam delapan kelas dengan nilai akurasi 96,05%, presisi 96,00%, sensitivitas 96,05%, dan F1 score 95,95%. Klasifier SVM Polinomial dengan kombinasi fitur geometris yang terdiri dari eccentricity, compactness, circularity, dan rasio sel berhasil mengklasifikasi sel pensil dengan nilai presisi 100,00%, sensitivitas 100,00%, dan F1 score 100,00%.
Mix histogram and gray level co-occurrence matrix to improve glaucoma prediction machine learning Jumanto, Jumanto; Nugraha, Faizal Widya; Harjoko, Agus; Muslim, Much Aziz; Alabid, Noralhuda N.
Journal of Soft Computing Exploration Vol. 4 No. 1 (2023): March 2023
Publisher : SHM Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joscex.v4i1.99

Abstract

Glaucoma is an eye disease that is the second leading cause of blindness. Examination of glaucoma by an ophthalmologist is usually done by observing the retinal image directly. Observations from one doctor to another may differ, depending on their educational background, experience, and psychological condition. Therefore, a glaucoma detection system based on digital image processing is needed. The detection or classification of glaucoma with digital image processing is strongly influenced by the feature extraction method, feature selection, and the type of features used. Many researchers have carried out various kinds of feature extraction for glaucoma detection systems whose accuracy needs to be improved. In general, there are two groups of features, namely morphological features and non-morphological features (image-based features). In this study, it is proposed to detect glaucoma using texture features, namely the GLCM feature extraction method, histograms, and the combined GLCM-histogram extraction method. The GLCM method uses 5 features and the Histogram uses 6 features. To distinguish between glaucoma and non-glaucoma eyes, the multi-layer perceptron (MLP) artificial neural network model serves as a classifier. The data used in this study consisted of 136 fundus images (66 normal images and 70 images affected by glaucoma). The performance obtained with this approach is an accuracy of 93.4%, a sensitivity of 86.6%, and a specificity of 100%.
Development of an Image Captioning Model to Assist The Activities of Visually Impaired Pedestrians in Urban Environments Sajid, Syahmi; Harjoko, Agus
JEPIN (Jurnal Edukasi dan Penelitian Informatika) Vol 11, No 2 (2025): Volume 11 No 2
Publisher : Program Studi Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26418/jp.v11i2.95754

Abstract

Visual impairment is a global issue with significant impacts on the mobility and safety of individuals, especially in urban environments. Artificial intelligence solutions, such as image captioning, promise assistance for people with visual impairments to aid their daily activities. However, the field of image captioning in this context still has performance limitations. To address this, this study proposes a hybrid method combining image feature extraction from VGG16, ResNet50, and YOLO on the encoder side with LSTM and BiGRU on the decoder side to generate descriptions that have proven to enhance model performance on the Flickr8k dataset in previous research. By adapting this method to the Visual Assistance dataset, incorporating image augmentation through a combination of rotation and zoom, and applying transfer learning to address the dataset size limitation, this study successfully improved the model"™s performance in supporting the activities of visually impaired pedestrians in urban environments. Evaluation results showed significant improvements in several evaluation metrics. Overall, this model shows improvement compared to previous research, where Sharma et al. (2022) reported that the InceptionV3-BiLSTM model with Adaptive Attention achieved a BLEU-4 score of only 0.266 on the Visual Assistance dataset. This study achieved a 60.53% increase in BLEU-4 score compared to previous research on the Visual Assistance dataset. Overall, this study provides a positive contribution to developing more effective and accurate solutions for visually impaired navigation users in urban environments.
Sistem Verifikasi Tanda Tangan Off-Line Berdasar Ciri Histogram Of Oriented Gradient (HOG) Dan Histogram Of Curvature (HoC) Widodo, Agus Wahyu; Harjoko, Agus
Jurnal Teknologi Informasi dan Ilmu Komputer Vol 2 No 1: April 2015
Publisher : Fakultas Ilmu Komputer, Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (334.075 KB) | DOI: 10.25126/jtiik.201521121

Abstract

Abstrak Tanda tangan dengan sifat uniknya merupakan salah satu dari sekian banyak atribut personal yang diterima secara luas untuk verifikasi indentitas seseorang, alat pembuktian kepemilikan berbagai transaksi atau dokumen di dalam masyarakat. Keberhasilan penggunaan ciri gradien dan curvature dalam bidang-bidang penelitian pengenalan pola dan bahwa tanda tangan dapat dikatakan merupakan hasil tulisan tangan yang tersusun atas beragam garis dan lengkungan (curve) yang memiliki arah atau orientasi merupakan alasan bahwa kedua ciri tersebut digunakan sebagai metoda verifikasi tanda tangan offline di penelitian ini. Berbagai implementasi dari pre-processing, ekstraksi dan representasi ciri, dan pembelajaran SVM serta usaha perbaikan yang telah dilakukan dalam penelitian ini menunjukkan hasil bahwa ciri HOG dan HoC mampu dimanfaatkan dalam proses verifikasi tanda tangan secara offline.  Pada basis data GPDS960Signature, HOG dan HoC yang dihitung pada ukuran sel 30 x 30 piksel memberikan dengan nilai %FRR terbaik 26,90 dan %FAR 37,56.  Sedangkan pada basis data FUM-PHSDB, HOG dn HoC yang dihitung pada ukuran 60 x 60 piksel memberikan nilai %FRR terbaik 4 dan %FAR 57. Kata kunci: verifikasi tanda tangan, curvature, orientation, gradient, histogram of curvature (HoC), histogram of oriented gradient (HOG) Abstract Signature with unique properties is one of the many personal attributes that are widely accepted to verify a person's identity, proof of ownership transactions instrument or document in the community. The successful use of gradient and curvature feature in the research fields of pattern recognition is the reason that both of these features are used as an offline signature verification method in this study. Various implementations of preprocessing, feature extraction and representation, and SVM learning has been done in the study showed results that HOG and HoC feature can be utilized in the process of offline signature verification.  HOG and HOC calculated on a combination of two different cell sizes at a time.  Improvement effort has been made and showed the expected results, although of little value. HOG and HOC calculated on a such cell sizes at a time. In database GPDS960Signature, best cell size is in 30 with the value 26.90% FRR and FAR 37.56%. While the database FUM-PHSDB, the best cell size is 60 with a value of 4% FRR and FAR 57%. Keywords: signature verification, curvature, orientation, gradient, a histogram of curvature (HOC), a histogram of oriented gradient (HOG)
Input Variable Selection for Oil Palm Plantation Productivity Prediction Model Suryotomo, Andiko Putro; Harjoko, Agus
Telematika Vol 20 No 1 (2023): Edisi Februari 2023
Publisher : Jurusan Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31315/telematika.v20i1.9674

Abstract

Purpose: This study aims to implement and improve a wrapper-type Input Variable Selection (IVS) to the prediction model of oil palm production utilizing oil palm expert knowledge criteria and distance-based data sensitivity criteria in order to measure cost-saving in laboratory leaf and soil sample testing.Methodology: The proposed approach consists of IVS process, searching the best prediction model based on the selected variables, and analyzing the cost-saving in laboratory leaf and soil sample testing.Findings/result: The proposed method managed to effectively choose 7 from 19 variables and achieve 81.47% saving from total laboratory sample testing cost.Value: This result has the potential to help small stakeholder oil palm planter to reduce the cost of laboratory testing without losing important information from their plantation.
KLASIFIKASI CITRA DENGAN POHON KEPUTUSAN Kusrini Kusrini; Sri Hartati; Retantyo Wardoyo; Agus Harjoko
JUTI: Jurnal Ilmiah Teknologi Informasi Vol 7, No 2, Juli 2008
Publisher : Department of Informatics, Institut Teknologi Sepuluh Nopember

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (613.649 KB) | DOI: 10.12962/j24068535.v7i2.a173

Abstract

Image classification can be done by using attribute of text that come along with the image, such as file name, size, or creator. Image classification also can be done base on visual content of the image. In this research, we implement a image classification model base on image visual content. The image classification is based on decision tree method that adapt C4.5 algorithm. The decision variable used in the decision tree generation process is image visual features, i.e. color moment order-1, color moment order-2, color moment order-3, entropy, energy, contrast, and homogeneity. The result of this research is an application that can classified image base on the knowledge of the previous classification cases.   Keywords: image classification, decision tree, C4.5 algorithm
Enhancing Deep Learning Model Using Whale Optimization Algorithm on Brain Tumor MRI Winarno, Winarno; Harjoko, Agus
Journal of Electronics, Electromedical Engineering, and Medical Informatics Vol 8 No 1 (2026): January
Publisher : Department of Electromedical Engineering, POLTEKKES KEMENKES SURABAYA

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35882/jeeemi.v8i1.941

Abstract

The increasing prevalence of brain cancer has emerged as a significant global health issue, with brain neoplasms, particularly gliomas, presenting considerable diagnostic and therapeutic obstacles. The timely and precise identification of such tumors is crucial for improving patient outcomes. This investigation explores the advancement of Convolutional Neural Networks (CNNs) for detecting brain tumors using MRI data, incorporating the Whale Optimization Algorithm (WOA) for the automated tuning of hyperparameters. Moreover, two callbacks, ReduceLROnPlateau and early stopping, were utilized to augment training efficacy and model resilience. The proposed model exhibited exceptional performance across all tumor categories. Specifically, the precision, recall, and F1-scores for Glioma were recorded as 0.997, 0.980, and 0.988, respectively; for meningioma, as 0.983, 0.986, and 0.984; for no tumors, as 0.998, 0.998, and 0.998; and for pituitary, as 0.997, 0.997, and 0.997. The mean performance metrics attained were 0.994 for precision, 0.990 for recall, and 0.992 for F1-score. The overall accuracy of the model was determined to be 0.991. Notably, incorporating callbacks within the CNN architecture improved accuracy to 0.994. Furthermore, when synergized with the WOA, the CNN-WOA model achieved a maximum accuracy of 0.996. This advancement highlights the effectiveness of integrating adaptive learning methodologies with metaheuristic optimization techniques. The findings suggest that the model sustains high classification accuracy across diverse tumor types and exhibits stability and robustness throughout training. The amalgamation of callbacks and the Whale Optimization Algorithm significantly bolster CNN performance in classifying brain tumors. These advancements contribute to the development of more reliable diagnostic instruments in medical imaging
Evaluation Of A Feature-Concatenated Model For Multiclass Diagnosis Of Pulmonary Diseases on An Imbalanced Dataset Ajitomo, Wahyu; Tyas, Dyah Aruming; Harjoko, Agus
International Journal of Artificial Intelligence Research Vol 9, No 2 (2025): December
Publisher : Universitas Dharma Wacana

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29099/ijair.v9i2.1519

Abstract

Lung diseases such as pneumonia, tuberculosis, and COVID-19 pose serious global health challenges, particularly in X-ray image classification where class distribution is often imbalanced. To address this issue, this study proposes a hybrid model based on concatenated CNN architectures and applies class weighting using focal loss multiclass. The dataset consists of 7,135 X-ray images divided into four main classes: pneumonia, tuberculosis, COVID-19, and normal. Focal loss with a gamma parameter of 2.0 is employed to enhance the model’s focus on minority classes. Evaluation results show that combined models such as DenseNet121 + VGG16 and VGG16 + ResNet50 achieve F1-scores of up to 0.87, outperforming single models. Grad-CAM visualizations also indicate that the combined models can recognize pathological areas more comprehensively and accurately. This approach proves effective in improving the accuracy and sensitivity of AI-based diagnostic systems.
Improved Wavelet-GLCM for Robust Batik Motif Classification Pradana, Gregorius Adi; Harjoko, Agus
IJCCS (Indonesian Journal of Computing and Cybernetics Systems) Vol 20, No 1 (2026): January
Publisher : IndoCEISS in colaboration with Universitas Gadjah Mada, Indonesia.

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22146/ijccs.108011

Abstract

Batik is a traditional Indonesian fabric made by applying wax on the fabric, then processed with a certain technique. The diversity of batik motifs often makes it difficult for people to recognize them. Therefore, a batik motif classification system is needed, one of the methods of which is based on digital image processing. However, in this method, variations in rotation and scale in the image often cause feature values to change, thus decreasing accuracy. To overcome this, this study proposes a robust Improved Wavelet-GLCM (IWGLCM) feature extraction method for classifying batik motif images that account for rotation and scale variations. This method combines the Gray Level Co-occurrence Matrix (GLCM) and statistical values from the results of the Discrete Wavelet Transform (DWT). These combined features are then classified using Support Vector Machine (SVM). In the test scenario with variations in rotation and scale at the same time, this method managed to achieve optimal performance with an accuracy of 95.00%, a precision of 95.08%, a recall of 95.00%, and an f1-score of 95.00%.