EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

DualDiscWaveGAN-Based Data Augmentation Scheme for Animal Sound Classification

Eunbeen Kim iD, Jaeuk Moon iD, Jonghwa Shim iD, Eenjun Hwang iD

Sensors · 10.02.2023 · 23 (4) · 2024

DOI: 10.3390/s23042024 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 10.02.2023. Verfasst von Eunbeen Kim und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Electrical Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea. Hauptthemen laut MEDLINE: Animals, Wild, Learning. Die Arbeit wurde bisher 17-mal zitiert (Crossref) und nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministry of Environment, Korea Ministry of Environment.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Animal sound classification (ASC) refers to the automatic identification of animal categories by sound, and is useful for monitoring rare or elusive wildlife. Thus far, deep-learning-based models have shown good performance in ASC when training data is sufficient, but suffer from severe performance degradation if not. Recently, generative adversarial networks (GANs) have shown the potential to solve this problem by generating virtual data. However, in a multi-class environment, existing GAN-based methods need to construct separate generative models for each class. Additionally, they only consider the waveform or spectrogram of sound, resulting in poor quality of the generated sound. To overcome these shortcomings, we propose a two-step sound augmentation scheme using a class-conditional GAN. First, common features are learned from all classes of animal sounds, and multiple classes of animal sounds are generated based on the features that consider both waveforms and spectrograms using class-conditional GAN. Second, we select data from the generated data based on the confidence of the pretrained ASC model to improve classification performance. Through experiments, we show that the proposed method improves the accuracy of the basic ASC model by up to 18.3%, which corresponds to a performance improvement of 13.4% compared to the second-best augmentation method.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Animals, WildLearningAnimalsModels, AnimalSound

Stichwörter der Autor:innen

GANanimal sound classificationdata augmentationdeep learning

Autor:innen und Institutionen

  • School of Electrical Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea.

ORCID-Profile: Eunbeen Kim, Jaeuk Moon, Jonghwa Shim, Eenjun Hwang

Förderung

  • Ministry of Environment
  • Korea Ministry of Environment

Zitieren

APA

Kim, E., Moon, J., Shim, J., & Hwang, E. (2023). DualDiscWaveGAN-Based Data Augmentation Scheme for Animal Sound Classification. Sensors, 23(4), 2024. https://doi.org/10.3390/s23042024

Vancouver

Kim E, Moon J, Shim J, Hwang E. DualDiscWaveGAN-Based Data Augmentation Scheme for Animal Sound Classification. Sensors. 2023;23(4):2024. doi:10.3390/s23042024

BibTeX
@article{Kim2023,
  author = {Kim, Eunbeen and Moon, Jaeuk and Shim, Jonghwa and Hwang, Eenjun},
  title = {DualDiscWaveGAN-Based Data Augmentation Scheme for Animal Sound Classification},
  journal = {Sensors},
  year = {2023},
  volume = {23},
  number = {4},
  pages = {2024},
  doi = {10.3390/s23042024}
}
Kostenlos starten →