Deep-learning-based Multi-behavior Classification of Animals for Efficient Health and Welfare Monitoring - Vol. 35 - No. 11

Oct 6, 2026

Uploaded by asba

Deep-learning-based animal behavior recognition is developed to support efficient health and welfare monitoring using an IoT-enabled surveillance camera. Outdoor images of a polar bear are processed by an object detection model to derive frame-wise locations, movement speed, and coordinates, which are then used to train a multi-behavior classification model. The system detects multiple behaviors, including stereotypical behavior, with strong accuracy (98.3% average precision at AP50 and 89.5% recognition accuracy) while remaining robust to outdoor noise.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages22

Reading Time55 min

AccessFree to download

Introduction

Background and motivation

Deep learning and IoT for behavior analysis

Study aims and contributions

Sensors and Materials, Vol. 35, No. 11 (2023) 3947–3968 3947 MYU Tokyo S & M 3461 Deep-learning-based Multi-behavior Classification of Animals for Efficient Health and Welfare Monitoring Ruqin Wang, 1* Wataru Noguchi,2 Koki Osada, 1 and Masahito Yamamoto3 1Graduate School of Information Science and Technology, Hokkaido University, Kita 14, Nishi 9, Kita-ku, Sapporo, Hokkaido 060-0814, Japan 2Education and Research Center for Mathematical and Data Science, Hokkaido University, Kita 12, Nishi 7, Kita-ku, Sapporo, Hokkaido 060-0812, Japan 3Faculty of Information Science and Technology, Hokkaido University, Kita 14, Nishi 9, Kita-ku, Sapporo, Hokkaido 060-0814, Japan (Received May 18, 2023; accepted September 12, 2023) Keywords: ethology, stereotypical behavior, animal behavior recognition, object detection, YOLOv5s With the development of sensor technologies, sensors have become increasingly embedded in various fields, becoming an indispensable part of our daily lives, research, and work. Notably, in ethology, surveillance cameras, a type of optical sensor, are extensively used alongside machine learning to analyze animal behaviors. However, simply feeding vast amounts of sensor data into servers for processing is neither efficient nor sustainable. In line with the prevailing trend towards edge computing, it is becoming increasingly important to process and integrate the captured sensor information directly within the sensor itself. While we have not fully achieved this, the application of deep learning methods to facilitate efficient and rapid processing with low computational demands is a necessary progression. In our study, we used a method for outdoor animal behavior analysis using multi-target classification, taking advantage of the potential efficiency gains provided by deep learning. We focus on a polar bear’s behaviors captured by an IoT-enabled surveillance camera in a zoo. The image data are first analyzed by using an object detection model to provide location sequences, movement speed, and coordinates of video frames, representing the animal’s state. Using these sensor data, we developed a classification model that accurately classifies multiple behaviors. The detection of these behaviors, including stereotypical behavior, illustrates the potential of our system to comprehensively monitor the animal health status. Our method achieved accurate detection [98.3% average precision (AP) 50] and multi-behavior recognition (accuracy of 89.5%), while maintaining robustness against outdoor noise. 1. Introduction The increasing integration of sensor technologies in various fields has become a defining trend in recent years, playing an integral role in research, industry, and daily life. Particularly in the domain of ethology, sensor technologies, notably surveillance cameras, are gaining * Corresponding author: e-mail: [ruqin.wang.q3@elms.hokudai.c.jp](ruqin.wang.q3@elms.hokudai.c.jp)[ ](ruqin.wang.q3@elms.hokudai.c.jp)[https://doi.org/10.18494/SAM4521](https://doi.org/10.18494/SAM4521) ISSN 0914-4935 © MYU K.K. [https://myukk.org/](https://myukk.org/) 3948 Sensors and Materials, Vol. 35, No. 11 (2023) momentum in the analysis of animal behaviors.(1) A continuous observation facilitated by IoT sensor technologies such as surveillance cameras provides extensive data on animal behavior, contributing significantly to fields such as ecology, wildlife conservation, animal welfare, and animal cognition research. Recent advances have made it possible to integrate sensors with sophisticated data processing methods such as machine learning, paving the way for the improved analysis of complex patterns and behaviors.(2) This trend aligns with the ongoing shift towards edge computing, where data captured by sensors is processed and integrated directly within the sensor itself, promoting efficiency and real-time processing.(3) As sensor technologies and deep learning techniques advance, more robust, accu

FAQs

What problem does the study address in animal welfare monitoring?

The study addresses inefficient processing of large sensor datasets by performing analysis on captured sensor information and enabling rapid, low-computation behavior monitoring for health and welfare.

How is the polar bear’s behavior recognition pipeline constructed?

The method uses object detection to obtain location sequences, movement speed, and coordinates from video frames, then feeds these sensor-derived data into a classification model to label multiple behaviors.

What performance and robustness results are reported?

The approach achieves accurate detection with 98.3% average precision at AP50 and multi-behavior recognition accuracy of 89.5%, while maintaining robustness under outdoor noise.

Oct 6, 2026

Uploaded by asba