Article Per Year (5 Year)
p-Index From 2020 - 2025
0.562
P-Index
This Author published in this journals
All Journal JURNAL MEDIA INFORMATIKA BUDIDARMA Journal of Computer System and Informatics (JoSYC)
Janson Hendryli
Universitas Tarumanagara, Jakarta
Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Electrical & Electronics Engineering Industrial & Manufacturing Engineering

Published : 3 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 3 Documents
Search

 1 
Performa Support Vector Machine Pada Klasifikasi Lahan dan Air Tanah Angellina Angellina; Dyah Erny Herwindiati; Janson Hendryli
JURNAL MEDIA INFORMATIKA BUDIDARMA Vol 7, No 1 (2023): Januari 2023
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/mib.v7i1.5279

Abstract

Groundwater is one of the sources of water in the world. The availability of groundwater is one of the factors that plays an important role in carrying out daily life activities, including for drinking, cooking, washing, irrigating rice fields, and many others. One source of groundwater in Jakarta is obtained from the Ciliwung River which is passed by the Bogor and Depok areas. However, the existence of springs and groundwater continues to decrease until now. The purpose of this paper is to discuss the first stage of the classification of groundwater availability in several sub-districts in the Bogor and Depok areas. The results of phase one will present a mapping of green areas along with their classification. Data taken from Landsat 8 Satellite Imagery - United States Geological Survey (USGS). The Support vector Machine (SVM) method is used to classify the availability of groundwater. The input data for the training process are the Normalized Difference Vegetation Index, Soil Adjusted Vegetation Index, and Enhanced Vegetation Index constants. The results of the evaluation using linear kernel produced a green F1 score of 89.58%, half green 65.62%, and dry 83.44%. While the results of the evaluation using the polynomial kernel produced a green F1 score of 83.58%, half green 25.68%, and dry 66.59%.
Performa Metode Convolutional Neural Network Pada Face Landmark Untuk Virtual Make Up Try On Dameethia Angeline; Erico Jochsen; Dyah Erny Herwindiati; Janson Hendryli
JURNAL MEDIA INFORMATIKA BUDIDARMA Vol 7, No 4 (2023): Oktober 2023
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/mib.v7i4.6619

Abstract

Make up or facial makeup, is an activity to change the appearance from its original form with the help of make up materials and tools. Make-up tools are beauty tools that are commonly used by most women to beautify the appearance of their faces with many shade choices. The shade on the make-up tool is the color usually used in make-up. Examples of make-up tools that are most often used include eyeshadow, blush on, and lipstick. These make-up tools are sold widely online and offline in physical stores. However, usually a tester is also needed so that those who want to buy can try the shade that suits them. When buying online, they often find it difficult to choose the right shade, while testers in physical stores are sometimes considered less hygienic because they have been used by many people. The aim of this paper is to measure the performance of the Convolutional Neural Network (CNN) method using the ResNet-50 architecture on facial landmarks for creating virtual make up try ons which can be an alternative to this problem. The facial image data source used is from the Kaggle site called Facial Keypoints Detection. The testing process produces 78.99% accuracy while the training process produces 95.12% accuracy. The evaluation results of this model use Root Mean Squared Error (RMSE) of 2.2577 and Mean Absolute Error (MAE) of 1.5389.
Pengenalan Bangunan Bersejarah Pura dengan Menggunakan Local Binary Pattern dan Support Vector Machine Erico Jochsen; Dameethia Angeline; Dyah Erny Herwindiati; Janson Hendryli
Journal of Computer System and Informatics (JoSYC) Vol 5 No 1 (2023): November 2023
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/josyc.v5i1.4553

Abstract

One area that has a rich cultural heritage is Bali. Bali is very well known as a very beautiful place and is often visited by tourists in Indonesia and outside Indonesia. Temple buildings in Bali have unique characteristics that reflect the richness of Indonesian culture. So many tourists are interested in vacationing there. However, due to the uniqueness of each temple building there, there is a lack of knowledge about the buildings being seen, so the main aim of this design is to develop a system for recognizing historical temple buildings in Indonesia through building images. More broadly, this design contribution can be applied in the development of similar systems for other historical regions in Indonesia, enriching efforts to preserve and promote cultural heritage nationally. Thus, this design not only paves the way for innovation in the field of image recognition, but also has a positive impact in preserving valuable cultural property. The method used for recognition is Local Binary Pattern as texture feature extraction from temple building images, while Support Vector Machine with a polynomial kernel is used to recognize temple buildings. It is hoped that the combination of these two methods can provide good results in recognizing temple buildings with the correct classification level. The accuracy of this design model using 90 percent training data and 10 percent test data was 45.93 percent, while when using 80 percent training data and 20 percent test data, the accuracy dropped slightly to 43.96 percent. When using 90 percent training data, the recognition of historical buildings produces a precision of 59 percent, a recall value of 71 percent, and an f1-score of 57 percent. On the other hand, with 80 percent training data, the recognition of historical buildings produces 62 percent precision, 72 percent recall value, and 57 percent f1-score.
Co-Authors Angellina Angellina Dameethia Angeline Dyah Erny Herwindiati Erico Jochsen