Article Per Year (5 Year)
p-Index From 2020 - 2025
0.408
P-Index
This Author published in this journals
All Journal AJIE (Asian Journal of Innovation and Entrepreneurship) JATI (Jurnal Mahasiswa Teknik Informatika)
Kusuma, Permadi
Unknown Affiliation
Computer Science & IT Economics, Econometrics & Finance

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Implementasi Steganografi Dengan Menggunakan Metode Masking And Filtering Untuk Menyisipkan Pesan Ke Dalam Spectrogram Audio: Indonesia Kusuma, Permadi; Prayudi, Yudi
Asian Journal of Innovation and Entrepreneurship Volume 09, Issue 01, January 2025
Publisher : UII

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.20885/ajie.vol9.iss1.art1

Abstract

When sending a message to a specific party and do not want the message to be known by other parties, it is important to avoid information leakage. However, the problem identified is that there is a lack of knowledge to detect audio Ste-ganography which requires technical methods that can read and view secret messages. One method that can be used in steganography is Masking and Filtering. Masking as a media marker on audio that can insert messages. Filtering gives value to the parts that have been given a mark.  This method is one that is often used because it is simple, fast in the data insertion process, and has a large enough storage capacity. The Masking and Filtering method is able to hide messages by inserting them into the audio Spectrogram as a storage medium. The filter is used to ensure that the hidden message is within the previously analyzed frequency range, thus making humans unable to clearly hear the additional audio that has been inserted which is the hidden message.  After the insertion is complete, the audio file is saved, and tests are performed to ensure that the audio quality is not compromised, and the hidden message remains undetected such as making modifications to the stego file to test the robustness and security of the hidden message. Based on research, steganography is difficult to detect by the naked eye, to retrieve messages that have been hidden, it can be done by displaying an audio Spectrogram that contains a secret message. How to see the hidden message using the Audacity application that can see sound waves. The result is that the message embedded in the audio is not damaged even though compression, cutting, and some of the processes carried out in the audio have been carried out.
ANALISIS FORENSIK DEEPFAKE BERBASIS CONVOLUTIONAL NEURAL NETWORK (CNN) UNTUK DETEKSI INKONSISTENSI TEKSTUR DAN POLA PADA CITRA WAJAH Raharjo, Toto; Irfan Adristi, Fikri; Yusroni Romadhona, Frendi; Rahmadi Syahputra, Rosi; Yusuf Halim, Muhammad; Ashshidiqie Rachman, Mikhail; Nur Marjianto, Rohsan; Santicho, Desylo; Kusuma, Permadi; Ramadhani, Erika
JATI (Jurnal Mahasiswa Teknik Informatika) Vol. 9 No. 2 (2025): JATI Vol. 9 No. 2
Publisher : Institut Teknologi Nasional Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.36040/jati.v9i2.13058

Abstract

Perkembangan teknologi artificial intelligence (AI) yang semakin pesat memunculkan teknologi deepfake yang menggunakan basis algoritma seperti generative adversarial networks (GANs) untuk memanipulasi citra wajah dengan tingkat realisme yang tinggi. Fenomena ini memunculkan tantangan dalam keamanan digital karena potensi penyalahgunaan, disinformasi, pelanggaran privasi, dan kejahatan dunia maya. Penelitian ini bertujuan mengembangkan model deteksi deepfake berbasis convolutional neural networks (CNN) yang mampu mengidentifikasi inkonsistensi tekstur dan pola biologis pada citra wajah secara efisien secara resource hadware dan efektif dalam proses deteksi image. Metode penelitian meliputi pengumpulan data dari dataset FaceForensics++ dan Celeb-DF, pra-proses data menggunakan augmentasi dan normalisasi, serta pengembangan model dengan arsitektur EfficientNetB0 yang dilatih menggunakan transfer learning pada dataset besar dengan menggunakan TensorFlow dan GPU. Hasil pengujian menunjukkan bahwa Model cenderung mengklasifikasikan gambar asli dengan nilai prediksi dari model neural network di atas 0,5, sementara gambar palsu justru dideteksi sebagai asli karena nilainya juga melebihi 0,5. Evaluasi menunjukkan bahwa model ini mencapai tingkat akurasi tinggi dalam mendeteksi manipulasi citra wajah, dengan teknik augmentasi yang meningkatkan kehandalan model terhadap berbagai skenario. Penelitian ini menyimpulkan bahwa pendekatan berbasis CNN efektif untuk deteksi deepfake, namun perlu pengembangan lebih lanjut untuk mengatasi keterbatasan pada data yang kompleks.
Co-Authors Ashshidiqie Rachman, Mikhail Irfan Adristi, Fikri Nur Marjianto, Rohsan Raharjo, Toto Rahmadi Syahputra, Rosi Ramadhani, Erika Santicho, Desylo Yudi prayudi Yusroni Romadhona, Frendi Yusuf Halim, Muhammad