The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
A classificação de espécies em sala de aula com base em sons de animais é uma tarefa altamente desafiadora, mesmo com a mais recente técnica de aprendizagem profunda aplicada. A dificuldade de distinguir as espécies é ainda maior quando o número de espécies é grande dentro da mesma classe. Este artigo apresenta uma nova abordagem para categorização precisa de espécies animais com base em seus sons, usando CNNs pré-treinados e um novo módulo de autoatenção adequado para sinais acústicos. O método proposto mostra-se eficaz, pois atinge uma precisão média de espécies de 98.37%. e a precisão mínima das espécies de 94.38%, a mais alta entre as linhas de base concorrentes, que incluem CNNs sem autoatenção e CNNs com CBAM, FAM e CFAM, mas sem pré-treinamento.
Kyungdeuk KO
Korea University
Jaihyun PARK
Korea University
David K. HAN
US Army Research Laboratory (ARL)
Hanseok KO
Korea University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Kyungdeuk KO, Jaihyun PARK, David K. HAN, Hanseok KO, "Channel and Frequency Attention Module for Diverse Animal Sound Classification" in IEICE TRANSACTIONS on Information,
vol. E102-D, no. 12, pp. 2615-2618, December 2019, doi: 10.1587/transinf.2019EDL8128.
Abstract: In-class species classification based on animal sounds is a highly challenging task even with the latest deep learning technique applied. The difficulty of distinguishing the species is further compounded when the number of species is large within the same class. This paper presents a novel approach for fine categorization of animal species based on their sounds by using pre-trained CNNs and a new self-attention module well-suited for acoustic signals The proposed method is shown effective as it achieves average species accuracy of 98.37% and the minimum species accuracy of 94.38%, the highest among the competing baselines, which include CNN's without self-attention and CNN's with CBAM, FAM, and CFAM but without pre-training.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019EDL8128/_p
Copiar
@ARTICLE{e102-d_12_2615,
author={Kyungdeuk KO, Jaihyun PARK, David K. HAN, Hanseok KO, },
journal={IEICE TRANSACTIONS on Information},
title={Channel and Frequency Attention Module for Diverse Animal Sound Classification},
year={2019},
volume={E102-D},
number={12},
pages={2615-2618},
abstract={In-class species classification based on animal sounds is a highly challenging task even with the latest deep learning technique applied. The difficulty of distinguishing the species is further compounded when the number of species is large within the same class. This paper presents a novel approach for fine categorization of animal species based on their sounds by using pre-trained CNNs and a new self-attention module well-suited for acoustic signals The proposed method is shown effective as it achieves average species accuracy of 98.37% and the minimum species accuracy of 94.38%, the highest among the competing baselines, which include CNN's without self-attention and CNN's with CBAM, FAM, and CFAM but without pre-training.},
keywords={},
doi={10.1587/transinf.2019EDL8128},
ISSN={1745-1361},
month={December},}
Copiar
TY - JOUR
TI - Channel and Frequency Attention Module for Diverse Animal Sound Classification
T2 - IEICE TRANSACTIONS on Information
SP - 2615
EP - 2618
AU - Kyungdeuk KO
AU - Jaihyun PARK
AU - David K. HAN
AU - Hanseok KO
PY - 2019
DO - 10.1587/transinf.2019EDL8128
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E102-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2019
AB - In-class species classification based on animal sounds is a highly challenging task even with the latest deep learning technique applied. The difficulty of distinguishing the species is further compounded when the number of species is large within the same class. This paper presents a novel approach for fine categorization of animal species based on their sounds by using pre-trained CNNs and a new self-attention module well-suited for acoustic signals The proposed method is shown effective as it achieves average species accuracy of 98.37% and the minimum species accuracy of 94.38%, the highest among the competing baselines, which include CNN's without self-attention and CNN's with CBAM, FAM, and CFAM but without pre-training.
ER -