Computer Engineering and Applications Journal (ComEngApp, Universitas Sriwijaya)
Not a member yet
102 research outputs found
Sort by
Classification of Atrial Fibrillation In ECG Signal Using Deep Learning
No full textAtrial fibrillation is a type of heart rhythm disorder that most often occurs in the world and can cause death. Atrial fibrillation can be diagnosed by reading an Electrocardiograph (ECG) recording, however, an ECG reading takes a long time and requires specialists to analyze the type of signal pattern. The use of deep learning to classify Atrial Fibrillation in ECG signals was chosen because deep learning has 10% higher performance compared to machine learning methods. In this research, an application for classification of Atrial Fibrillation was developed using the 1- Dimentional Convolutional Neural Network (CNN 1D) method. There are 6 configurations of the 1D CNN model that were developed by varying the configuration on the learning rate and batch size. The best model obtained 100% accuracy, 100% precision, 100% recall, and 100% F1 Score
Littering Activities Monitoring using Image Processing
No full textLittering is a human behavior that become a habit since childhood. Even though there are rules that prohibit this behavior, the community still continues to do so. In order to limit this bad behavior, a device that can monitor and provide notifications is needed. In this research, proposed device can identify human activities by utilizing webcam-based image processing. It is processed by machine learning using the Recurrent Neural Network (RNN). The monitoring device produced in this research works by comparing the captured image data with dataset. The captured image data are extracted into figures and form several coordinate points on the human body. Then, the system classifies the human activities into two categories, i.e., normal or littering. This device will provide an output in the form of a ewarning every time the activity of littering is detected
Robot Vision Pattern Recognition of the Eye and Nose Using the Local Binary Pattern Histogram Method
No full textThe local binary pattern histogram (LBPH) algorithm is a computer technique that can detect a person\u27s face based on information stored in a database (trained model). In this research, the LBPH approach is applied for face recognition combined with the embedded platform on the actuator system. This application will be incorporated into the robot\u27s control and processing center, which consists of a Raspberry Pi and Arduino board. The robot will be equipped with a program that can identify and recognize a human\u27s face based on information from the person\u27s eyes and nose. Based on the results of facial feature identification testing, the eyes were recognized 131 times (87.33%), and the nose 133 times (88.67%) out of 150 image data samples. From the test results, an accuracy rate of 88%, the partition rate of 95.23%, the recall of 30%, the specificity of 99%, and the F1-Score of 57.5% were obtained
Identification of Indonesian Authors Using Deep Neural Networks
No full textAuthor Name Disambiguation (AND) is a problem that occurs when a set of publications contains ambiguous names of authors, i.e. the same author may appear with different names (synonyms) in other published papers, or author (authors) who may be different who may have the same name (homonym). In this final project, we will design a model with a Deep Neural Network (DNN) classifier. The dataset used in this final project uses primary data sourced from the Scopus website. This research focuses on integrating data from Indonesian authors. Parameters accuracy, sensitivity and precision are standard benchmarks to determine the performance of the method used to solve AND problems. The best DNN classification model achieves 99.9936% Accuracy, 93.1433% Sensitivity, 94.3733% Precision. Then for the highest performance measurement, the case of Non Synonym-Homonym (SH) has 99.9967% Accuracy, 96.7388% Sensitivity, and 97.5102% Precision
Deforestation Analysis in Taba Penanjung District with NDVI
No full textForests are cleared to expand residential areas and plantations or change the allocation of forest land to non-forest (deforestation). This study aims to create a map of the condition of the forest area and find out the areas affected by deforestation and determine the rate of change in the forest area in the Taba Penanjung District forest area. First data on Landsat 8 images geometric correction performed to position image data so that it matches the actual coordinates and performing Radiometric Correction is used to correct if an error or distortion occurs due to imperfect operation and sensors. NDVI Method is the method used for comparing the greenness of vegetation in satellite imagery, which uses band 4 (Red) and band 5 (NIR), which is processed by ArcMap software. The results of this study produced a map of the condition of forest areas and the area of land affected by deforestation. Forest turn-over rates were described in the annual trend from 2013 to 2018. Furthermore, this research shows that deforestation in the Taba Penanjung district has happened in 58% of the total area of 23, 747 ha. Although the deforestation has decreasing value in 2015 by 1.6%, it showed that there were increasing values in deforestation rate in 2014, 206, 2017 by 1.4%, 1.9%, and 9.5% respectively from the total area of 23, 747 ha
Multiclass Segmentation of Pulmonary Diseases using Convolutional Neural Network
No full textPulmonary disease has affected tens of millions of people in the world. This disease has also become the cause of death of millions of its sufferers every year. In addition, lung disease has also become the cause of other respiratory complications, which also causes the death of the sufferer. The diagnosis of pulmonary diseases through medical imaging is a significant challenge in computer vision and medical image processing. The difficulty is due to the wide variety in infected areas\u27 shape, dimension, and location. Another challenge is to differentiate one lung disease from the other. Discriminating pulmonary diseases is a notable concern in the diagnosis of pulmonary disease. We have adopted the deep learning convolutional neural network in this study to address these challenges. Seven models were constructed using the Mask Region-based Convolutional Neural Network (Mask-RCNN) architecture to detect and segment infected areas within the lung region from CT scan imagery. The evaluation results show that the best model obtained scores of 91.98%, 85.25%, and 93.75% for DSC, MIoU, and mAP, respectively. The segmentation results are then visualized
Performance Comparison of Feature Face Detection Algorithm on The Embedded Platform
No full textThe intensity of light will greatly affect every process carried out in image processing, especially facial images. It is important to analyze how the performance of each face detection method when tested at several lighting levels. In face detection, various methods can be used and have been tested. The FLP method automates the identification of the location of facial points. The Fisherface method reduces the dimensions obtained from PCA calculations. The LBPH method converts the texture of a face image into a binary value, while the WNNs method uses RAM to process image data, using the WiSARD architecture. This study proposes a technique for testing the effect of light on the performance of face detection methods, on an embedded platform. The highest accuracy was achieved by the LBPH and WNNs methods with an accuracy value of 98% at a lighting level of 400 lx. Meanwhile, at the lowest lighting level of 175 lx, all methods have a fairly good level of accuracy, which is between 75% to 83%
Detection of Diabetic Retinopathy Using Convolutional Neural Network (CNN)
No full textOne of the complications of Diabetes Mellitus, namely Diabetic Retinopathy (DR) damages the retina of the eye and has five levels of severity: Normal, Mild, Medium, Severe and Proliferate. If not detected and treated, this complication can lead to blindness. Detection and classification of this disease is still done manually by an ophthalmologist using an image of the patient\u27s eye fundus. Manual detection has the disadvantage that it requires an expert in the field and the process is difficult. This research was conducted by detecting and classifying DR disease using Convolutional Neural Network (CNN). The CNN model was built based on the VGG-16 architecture to study the characteristics of the eye fundus images of DR patients. The model was trained using 4750 images which were rescaled to 256 X 256 size and converted to grayscale using the BT-709 (HDTV) method. The CNN-based software with VGG- 16 architecture developed resulted in an accuracy of 62% for the detection and classification of 100 test images based on five DR severity classes. This software produces the highest Sensitivity value in the Normal class at 90% and the largest Specificity value in the Mild class at 97.5%
Performance Evaluation on Applied Low-Cost Multi-Sensor Technology in Air Pollution Monitoring
No full textThis research aims to discuss the application of multi-sensor network technology for the monitoring of indoor air pollution. Indoor air pollution has become a severe problem that affects public health, especially indoor parking. The indoor air pollution monitoring system will provide information about vehicle exhaust emission levels. We have improved the system to identify six parameters of the vehicles\u27 gas emissions within a different location at once. This research aimed to measure the parameter of Carbon Monoxide (CO), Carbon Dioxide (CO2), Hydro Carbon (HC), temperature and humidity, and levels of particulates in the air (PM10). The performance of this system shows good ability to compare the results of measurements of air quality measuring professionals. In this study, we investigated the performance of a custombuilt prototype developed under the android-based application to detect air pollution levels in the parking area. Our objective was to evaluate the suitability of a low-cost multi-sensor network for monitoring air pollution in parking and the other area. The benefit of our approach is that its time and space complexity make it valuable and efficient for real-time monitoring of air pollution
Leukaemia Identification based on Texture Analysis of Microscopic Peripheral Blood Images using Feed-Forward Neural Network
No full textLeukaemia is very dangerous because it includes liquid tumour that it cannot be seen physically and is difficult to detect. Alternative detection of Leukaemia using microscopy can be processed using a computing system. Leukemia disease can be detected by microscopic examination. Microscopic test results can be processed using machine learning for classification systems. The classification system can be obtained using Feed-Forward Neural Network. Extreme Learning Machine (ELM) is a neural network that has a feedforward structure with a single hidden layer. ELM chooses the input weight and hidden neuron bias at random to minimize training time based on the Moore Penrose Pseudoinverse theory. The classification of Leukaemia is based on microscopic peripheral blood images using ELM. The classification stages consist of pre-processing, feature extraction using GLRLM, and classification using ELM. This system is used to classify Leukaemia into three classes, that is acute lymphoblastic Leukaemia, chronic lymphoblastic Leukaemia, and not Leukaemia. The best results were obtained in ten hidden nodes with an accuracy of 100%, a precision of 100%, a withdrawal of 100%