Title: Research on feeding behaviour of fish by using spatial and temporal features of depth images

Authors: Lvqing Bi; Ping Huang; Tianlin Huang; Zhixun Liang; Donghui Guo; Jincun Zheng

Addresses: School of Electronic Science and Engineering (National Model Microelectronics College), Xiamen University, Xiamen, China; School of Physics and Telecommunications Engineering, Guangxi Colleges and Universities Key Laboratory of Complex System Optimization and Big Data Processing, Research Center for Intelligent Information and Communication Technology, Yulin Normal University, Yulin, China ' School of Physics and Telecommunications Engineering, Yulin Normal University, Yulin, China ' School of Physics and Telecommunications Engineering, Yulin Normal University, Yulin, China ' School of Big Date and Computer, Hechi University, Yizhou, China ' School of Electronic Science and Engineering (National Model Microelectronics College), Xiamen University, Xiamen, China ' School of Physics and Telecommunications Engineering, Guangxi Colleges and Universities Key Laboratory of Complex System Optimization and Big Data Processing, Yulin Normal University, Yulin, China

Abstract: A novel method for evaluating the feeding behaviour of fish based on spatial-temporal features of the near-infrared depth image is introduced in this paper. First, the continuous feeding depth image in a specific spatial layer was collected. Then, the change of the pixels in adjacent frames of the depth image is calculated by using the frame difference method, which is used as the evaluation of the feeding intensity at that time. After that, all the evaluation values during the feeding period are sequentially composed into a sequence to obtain the time-domain sequence data. Finally, the bidirectional gated recurrent units (Bi-GRU) are employed to identify the feeding state of fish. The proposed method performs well in a complex breeding environment, with an average identification accuracy of 98.54%, and requires fewer computational resources, so it can be applied to the real-time intelligent feeding system.

Keywords: feeding states classification; spatial-temporal characteristics; Bi-GRU; near-infrared depth image.

DOI: 10.1504/IJBIC.2024.136724

International Journal of Bio-Inspired Computation, 2024 Vol.23 No.2, pp.125 - 134

Received: 10 Aug 2022
Accepted: 02 Aug 2023

Published online: 19 Feb 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article