Drone Classification with a Hybrid Deep Learning Approach Based on  Mel-Spectrogram Representation

Nida Kumbasar; Rabiye Kılıç; Emin Argun Oral; Yücel Özbek

Makale detayları

Drone Classification with a Hybrid Deep Learning Approach Based on Mel-Spectrogram Representation

Yazarlar:

Nida Kumbasar | Rabiye Kılıç | Emin Argun Oral | Yücel Özbek

Yayıncı:

Van Yuzuncu Yıl University

Ana Başlıklar:

Sinyal İşleme

Dergi:

Yüzüncü Yıl Üniversitesi Fen Bilimleri Enstitüsü Dergisi Vol.30 Issue 2, pp.608-620

Belge Tipi:

Makale

Yayın Tarihi:

8/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001357512

pISSN:

1300-5413

Anahtar kelimeler:

Özet:

 With technological advancements, the use of drones has become increasingly widespread in both civilian and military sectors in recent years. There is a need for technologies that can detect and identify the presence, type, or flight mode of a drone with remote sensing signals in situations that pose a security threat. This study is based on the classification of radio frequency RF signals from various drones under different flight modes using Mel-spectrogram representations. Within the scope of the study, the VGG19 deep learning model is used as both a classifier and a feature extractor for the SVM classifier. On the other hand, the study compares the performance of RF signals in low and high frequency bands separately and in concatenated versions. In the results obtained in the experimental studies, the VGG19SVM hybrid model showed the highest performance over the Mel-spectrogram of the concatenated low and high LH frequencies. The accuracy performances were 100 in the 2-Class problem where drone presence was detected drone -no drone, 90.78 in the 4-Class problem where drone types were classification Bebop-AR-Phantom-no drone, and 86.6 in the 10-Class problem where drone modes were obtained.

 With technological advancements, the use of drones has become increasingly widespread in both civilian and military sectors in recent years. There is a need for technologies that can detect and identify the presence, type, or flight mode of a drone with remote sensing signals in situations that pose a security threat. This study is based on the classification of radio frequency RF signals from various drones under different flight modes using Mel-spectrogram representations. Within the scope of the study, the VGG19 deep learning model is used as both a classifier and a feature extractor for the SVM classifier. On the other hand, the study compares the performance of RF signals in low and high frequency bands separately and in concatenated versions. In the results obtained in the experimental studies, the VGG19SVM hybrid model showed the highest performance over the Mel-spectrogram of the concatenated low and high LH frequencies. The accuracy performances were 100 in the 2-Class problem where drone presence was detected drone -no drone, 90.78 in the 4-Class problem where drone types were classification Bebop-AR-Phantom-no drone, and 86.6 in the 10-Class problem where drone modes were obtained.

 With technological advancements, the use of drones has become increasingly widespread in both civilian and military sectors in recent years. There is a need for technologies that can detect and identify the presence, type, or flight mode of a drone with remote sensing signals in situations that pose a security threat. This study is based on the classification of radio frequency RF signals from various drones under different flight modes using Mel-spectrogram representations. Within the scope of the study, the VGG19 deep learning model is used as both a classifier and a feature extractor for the SVM classifier. On the other hand, the study compares the performance of RF signals in low and high frequency bands separately and in concatenated versions. In the results obtained in the experimental studies, the VGG19SVM hybrid model showed the highest performance over the Mel-spectrogram of the concatenated low and high LH frequencies. The accuracy performances were 100 in the 2-Class problem where drone presence was detected drone -no drone, 90.78 in the 4-Class problem where drone types were classification Bebop-AR-Phantom-no drone, and 86.6 in the 10-Class problem where drone modes were obtained.

 Teknolojik gelişmelerle birlikte son yıllarda hem sivil hem de askeri sektörde dron kullanımı giderek yaygınlaşmaktadır. Güvenlik tehdidi oluşturan durumlarda uzaktan algılama sinyalleriyle bir dronun varlığını, türünü veya uçuş modunu tespit edip tanımlayabilen teknolojilere ihtiyaç duyulmaktadır. Bu çalışma, Mel-spektrogram gösterimleri kullanılarak farklı uçuş modları altında çeşitli dronlardan gelen radyo frekans RF sinyallerinin sınıflandırılmasına dayanmaktadır. Çalışma kapsamında, VGG19 derin öğrenme modeli hem bir sınıflandırıcı hem de SVM sınıflandırıcısı için bir özellik çıkarıcı olarak kullanılmıştır. Öte yandan çalışma, RF sinyallerinin düşük ve yüksek frekans bantlarındaki performansını ayrı ayrı ve birleştirilmiş versiyonlarda karşılaştırmaktadır. Deneysel çalışmalarda elde edilen sonuçlarda, VGG19SVM hibrit modeli, birleştirilmiş düşük ve yüksek LH frekansların Mel-spektrogramı üzerinde en yüksek performansı göstermiştir. dron varlığının tespit edildiği 2-Sınıf problemde dron var - dron yok doğruluk performansları 100, dron tiplerinin sınıflandırıldığı 4-Sınıf problemde Bebop-AR-Phantom-dron yok 90.78 ve dron modlarının elde edildiği 10-Sınıf problemde ise 86.6 olarak gerçekleşmiştir.

Makale detayları

Drone Classification with a Hybrid Deep Learning Approach Based on Mel-Spectrogram Representation

Yazarlar:

Nida Kumbasar | Rabiye Kılıç | Emin Argun Oral | Yücel Özbek

Yayıncı:

Van Yuzuncu Yıl University

Ana Başlıklar:

Sinyal İşleme

Dergi:

Yüzüncü Yıl Üniversitesi Fen Bilimleri Enstitüsü Dergisi Vol.30 Issue 2, pp.608-620

Belge Tipi:

Makale

Yayın Tarihi:

8/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001357512

pISSN:

1300-5413

Drone classification

Hybrid model

Mel-spectrogram

RF signalDrone sınıflandırma

Hibrit model

Mel-spektrogram

RF sinyal

Sayı

Makale detayları

Yazarlar: Nida Kumbasar | Rabiye Kılıç | Emin Argun Oral | Yücel Özbek

Yayıncı: Van Yuzuncu Yıl University

Ana Başlıklar: Sinyal İşleme

Dergi: Yüzüncü Yıl Üniversitesi Fen Bilimleri Enstitüsü Dergisi Vol.30 Issue 2, pp.608-620

Belge Tipi: Makale

Yayın Tarihi: 8/2025

Belge Dili: İngilizce

Document ID (Org No.): 2001357512

pISSN: 1300-5413

Sayı

Yazarlar:

Nida Kumbasar | Rabiye Kılıç | Emin Argun Oral | Yücel Özbek

Yayıncı:

Van Yuzuncu Yıl University

Ana Başlıklar:

Sinyal İşleme

Dergi:

Yüzüncü Yıl Üniversitesi Fen Bilimleri Enstitüsü Dergisi Vol.30 Issue 2, pp.608-620

Belge Tipi:

Makale

Yayın Tarihi:

8/2025

Belge Dili:

İngilizce

Document ID (Org No.):

2001357512

pISSN:

1300-5413