Advanced Computer Vision Methods for Tracking Wild Birds from Drone Footage

Author:

Mpouziotas Dimitris1ORCID,Karvelis Petros1ORCID,Stylios Chrysostomos2ORCID

Affiliation:

1. Department of Informatics and Telecommunications, University of Ioannina, 451 10 Ioannina, Greece

2. Industrial Systems Insitute, Athena RC, 263 31 Patra, Greece

Abstract

Wildlife conservationists have historically depended on manual methods for the identification and tracking of avian species, to monitor population dynamics and discern potential threats. Nonetheless, many of these techniques present inherent challenges and time constraints. With the advancement in computer vision techniques, automated bird detection and recognition have become possible. This study aimed to further advance the task of detecting wild birds using computer vision methods with drone footage, as well as entirely automating the process of detection and tracking. However, detecting objects from drone footage presents a significant challenge, due to the elevated altitudes, as well as the dynamic movement of both the drone and the birds. In this study, we developed and introduce a state-of-the-art model titled ORACLE (optimized rigorous advanced cutting-edge model for leveraging protection to ecosystems). ORACLE aims to facilitate robust communication across multiple models, with the goal of data retrieval, rigorously using various computer vision techniques such as object detection and multi-object tracking (MOT). The results of ORACLE’s vision models were evaluated at 91.89% mAP at 50% IoU.

Publisher

MDPI AG

Reference44 articles.

1. Paul, P.K., Choudhury, A., Biswas, A., and Singh, B.K. (2022). Drone Applications in Wildlife Research—A Synoptic Review. Environmental Informatics: Challenges and Solutions, Springer Nature Singapore.

2. Possibility of applying unmanned aerial vehicle (UAV) and mapping software for the monitoring of waterbirds and their habitats;Han;J. Ecol. Environ.,2017

3. Environmental variations in a semi-enclosed embayment (Amvrakikos Gulf, Greece)–reconstructions based on benthic foraminifera abundance and lipid biomarker pattern;Naeher;Biogeosciences,2012

4. Building of an edge enabled drone network ecosystem for bird species identification;Das;Ecol. Inform.,2021

5. The breeding biology of the Dalmatian Pelican Pelecanus crispus;Crivelli;Ibis,1998

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3