Claim Missing Document
Check
Articles

Found 5 Documents
Search

Evaluasi Metode Ekstraksi Fitur Hu Moment Invariants untuk Pengenalan Aktivitas Manusia Kurniawan, Hans Christian; Soemarto, Kevin Suryajaya; Yahya, Bernardo Nugroho
Jurnal Telematika Vol. 15 No. 2 (2020)
Publisher : Yayasan Petra Harapan Bangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61769/telematika.v15i2.367

Abstract

Vision-based Human Activity Recognition has been widely used due to a bunch of video data availability in the present days through CCTV and another mechanism which contains some human activities. This data is going to be very useful to improve and automate decision-making in several fields including security surveillance. In this field, it is important to achieve a good performance (i.e., accuracy) inefficient computational time. While there are many approaches in this field, most complex approaches require high computational time. In this work, we are evaluating Hu Moments performance, as well as being compared to other methods (i.e., Zernike Moment and Histogram of Oriented Gradient) by its accuracy and computational time. We also improved HAR flow by adding image denoising which has proven effective in increasing accuracy. The testing process includes videos that contain human activities such as walking, jogging, and running. The result shows that Hu Moments is superior among other methods, however there’s also some room for improvements found through this experiment.  Dalam era di mana terdapat banyak data video yang berisi aktivitas manusia, baik melalui rekaman CCTV maupun mekanisme lain, data tersebut menjadi sangat berharga untuk dapat diproses untuk pengenalan aktivitas manusia, atau Human Activity Recognition (HAR) yang dapat membantu pengambilan keputusan, di antaranya security surveillance. Untuk itu, diperlukan akurasi yang tinggi dan waktu komputasi yang efisien. Meskipun telah banyak metode di ranah ini, suatu teknik yang kompleks pada umumnya membutuhkan waktu komputasi yang tinggi. Dalam penelitian ini, dilakukan evaluasi dengan menggunakan metode Hu Moments yang akan dibandingkan dengan metode lainnya, yaitu Zernike Moment dan Histogram of Oriented Gradient (HOG), untuk segi akurasi dan waktu komputasinya. Ditambahkan juga tahap image denoising yang mampu meningkatkan akurasi. Proses pengujian menggunakan berbagai data video aktivitas manusia yang meliputi: berjalan, joging, dan berlari. Hasil riset menunjukkan bahwa metode Hu Moments memiliki performa yang lebih unggul dibandingkan metode ekstraksi fitur lainnya. Berdasarkan eksperimen yang dilakukan, terdapat beberapa area yang masih dapat ditingkatkan, untuk penelitian selanjutnya.
Penerapan Convolutional Neural Network untuk Melakukan Estimasi Pitch pada Rekaman Suara Penyanyi Pratama, Dionisius; Heryanto, Hery; Kurniawan, Hans Christian
Jurnal Telematika Vol. 16 No. 2 (2021)
Publisher : Yayasan Petra Harapan Bangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61769/telematika.v16i2.396

Abstract

A musical performance is determined by the intonation accuracy, which is the pitch accuracy of a musician or musical instrument, whether a tone is played 'in tune' or not. Therefore, to determine the intonation quality of a musical performance, it is necessary to estimate the pitch. In this research, a one-dimensional Convolutional Neural Network (CNN) is used to estimate the pitch from singing voice recording. After pitch estimation, Dynamic Time Warping (DTW) method is used to calculate the similarity (measured in distance) of pitch estimation results with the recording template from the dataset to determine intonation accuracy. This research uses several preprocessing methods, such as quantization pitch label, spectrogram generation, scaling, and spectrogram recoloring. The CNN method for performing pitch estimation is tested using five songs from the MIR-QBSH dataset. CNN testing is done by applying four architectural designs by combining epoch values, learning rate, number of filters in each convolutional layer, and number of convolutions to find the best combination that produces the highest accuracy. Based on the test results, the model built can produce the highest average accuracy of 97.425% with a difference between the average accuracy and the average validation accuracy of 14.383%. The optimal threshold value for distance is in the range of 1000-1500.  Pembawaan karya musik yang baik ditentukan dari ketepatan intonasi yang merupakan akurasi pitch dari sebuah nada yang dikeluarkan oleh seorang musisi atau instrumen musik, diproduksi dengan tepat atau tidak. Maka dari itu, untuk menentukan kualitas intonasi penampilan suatu karya musik, estimasi pitch perlu dilakukan. Pada penelitian ini, sebuah Convolutional Neural Network (CNN) satu dimensi digunakan untuk melakukan estimasi pitch dari rekaman suara nyanyian. Setelah estimasi pitch dilakukan, maka digunakan metode Dynamic Time Warping (DTW) untuk melakukan pengujian kemiripan (dalam distance) hasil estimasi pitch dengan template rekaman dari dataset. Pengujian tersebut dilakukan untuk menentukan ketepatan intonasi. Beberapa metode preprocessing yang dilakukan adalah pembulatan pitch label, pembuatan spektogram, scaling, dan pewarnaan ulang spektogram. Metode CNN untuk melakukan estimasi pitch diuji dengan menggunakan lima lagu dari dataset MIR-QBSH. Pengujian CNN dilakukan dengan menerapkan empat rancangan arsitektur dengan mengombinasikan nilai epoch, learning rate, jumlah filter pada setiap convolutional layer, dan jumlah konvolusi untuk mencari kombinasi terbaik yang menghasilkan akurasi tertinggi. Berdasarkan hasil pengujian, model yang dibangun dapat menghasilkan rata-rata akurasi tertinggi sebesar 97,425% dengan selisih antara rata-rata akurasi dan rata-rata akurasi validasi sebesar 14,383%. Nilai threshold yang optimal untuk distance berada pada rentang 1000-1500.
Analisis Pengaruh Design Pattern Terhadap Pemeliharaan Perangkat Lunak Learning Management System Kevin, Albertus; Kurniawan, Hans Christian
Jurnal Telematika Vol. 18 No. 1 (2023)
Publisher : Yayasan Petra Harapan Bangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61769/telematika.v18i1.543

Abstract

Software development focuses more on functionality so code quality is neglected. Design patterns can support faster development while supporting good code quality thus affecting long-term software maintenance. This research focuses on analyzing the effect of design patterns on software maintenance with open-source learning management system software in terms of design pattern characteristics. Five design patterns were applied and analyzed to assess the implementation method and design pattern characteristics. Testing is done using PHP Metrics. There are nine metrics to measure the complexity, size, cohesion, and dependability of each class. Overall, the design pattern had a good impact. Method templates and mediators have a good impact on cohesion, size, dependability, and complexity. Singletons increased the number of classes. Builders and strategies don't have much impact on size and complexity. Each design pattern generates a new class, so the complexity, dependencies, and size of the code are abstracted into that class.
Perbandingan Penerapan Relational Database Dan Graph Database Dalam Sistem Rekomendasi Film Florentina, Jennifer; Kurniawan, Hans Christian
Jurnal Telematika Vol. 18 No. 2 (2023)
Publisher : Yayasan Petra Harapan Bangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61769/telematika.v18i2.608

Abstract

Recommendation systems are used in various applications, such as e-commerce, social media, and others in building recommendation systems that require databases as data storage. The importance of database selection on system performance has increased research on the application of various types of databases in recommendation systems, including this research. This research compares latency and memory usage between relational databases and graph databases in movie recommendation systems. The main indicators in this research are the threshold value for the similarity value limit and the recommendation system technique used. There are 3 techniques used, namely content-based filtering using Jaccard similarity, collaborative filtering using cosine similarity, and hybrid filtering which is a combination of content-based filtering and collaborative filtering. The database used is PostgreSQL for relational databases and Neo4j for graph databases. Based on testing at various threshold values, the latency and memory usage values of the two databases are compared. In the content-based filtering technique, PostgreSQL has a latency time of 120-150 seconds and memory usage of 119-120 MB, while Neo4j is 6-7 seconds and 41-43 MB. In the collaborative filtering technique, PostgreSQL has a latency time of 3-4 seconds and memory usage of 119 - 120 MB, while Neo4j is 4-5 seconds and 24 - 26 MB. In the hybrid filtering technique, PostgreSQL has a latency time of 3-4 seconds and a memory usage of 24 - 26 MB. In the hybrid filtering technique, PostgreSQL has a latency time of 125-150 seconds and memory usage of 119-120 MB, while Neo4j has 9-11 seconds and 32-34 MB.
Does INDONESIAN CAPITAL MARKET EFFICIENT?: A RELATION BETWEEN PRICE-VOLUME Asnawi, Said Kelana; Pratama, Samuel; Kurniawan, Hans Christian; Rodjana, Samuel Yosua
Jurnal Ilmiah Ekonomi Dan Bisnis Vol. 20 No. 2 (2023)
Publisher : Universitas Lancang Kuning

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31849/jieb.v20i2.13019

Abstract

Efficient markets show prices have reflected information. In an efficient market, the pattern of price movements is a random walk, meaning that prices cannot be predicted accurately, so investors do not get abnormal returns. The informations used in this study are: lag-return (r-1); lag return(r-2); trading volume, as well as the synergy between (r-1) and trading volume. This research found that the coefficient was not significant in almost all tests. Investors cannot use past information to get abnormal returns Thus the efficient market hypothesis is proven. This efficient market situation shows that all market participants have equal opportunities in terms of risk-return.