Feature fusion strategy and improved GhostNet for accurate recognition of fish feeding behavior

•An LC-GhostNet lightweight network based on the GhostNet is proposed.•Fusion of Mel spectrogram, STFT feature map, and CQT feature map.•The method can quantify the feeding behavior of fish into four classes.•The feature fusion accuracy reaches 97.941%, which is higher than a single feature. In aqua...

Full description

Bibliographic Details

Published in:	Computers and electronics in agriculture Vol. 214; p. 108310
Main Authors:	Du, Zhuangzhuang, Xu, Xianbao, Bai, Zhuangzhuang, Liu, Xiaohang, Hu, Yang, Li, Wanchao, Wang, Cong, Li, Daoliang
Format:	Journal Article
Language:	English
Published:	Elsevier B.V 01.11.2023
Subjects:	agriculture aquaculture data collection electronics Feature fusion fish Fish feeding intensity LC-GhostNet lightweight network nondestructive methods On-demand feeding system Oplegnathus LC-GhostNet lightweight network Feature fusion Fish feeding intensity On-demand feeding system
ISSN:	0168-1699, 1872-7107
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!