|国家科技期刊平台
首页|期刊导航|北京测绘|基于深度学习的野生动物图像识别研究综述

基于深度学习的野生动物图像识别研究综述OA

Review of deep learning-based wildlife image recognition studies

中文摘要英文摘要

国家对生态文明建设重视程度的不断加深及计算机能力的重大突破,为实现更高效准确的野生动物图像识别提供了新的契机,基于计算机视觉的深度学习(DL)技术在图像识别领域发挥出巨大的优势.将深度学习算法应用于野生动物图像识别中可以捕捉到更加细致准确的野生动物信息,可以更好地帮助管理者对野生动物进行识别与监测,保护生态环境与物种多样性.本文从公开数据集与野外数据采集两方面入手,剖析了深度学习的研究现状,介绍了深度学习算法在野生动物图像识别上的研究进展,重点介绍区域卷积神经网络(R-CNN)、YOLO系列算法的现状,旨在为更高效的野生动物图像识别提供理论依据,为图像识别提供新的思路.

The deepening of national attention to the construction of ecological civilization and the major breakthrough of computer ability provide a new opportunity for realizing more efficient and accurate wildlife image recognition.The deep learning(DL)technology based on computer vision has played a great advantage in the field of image recognition.The application of the DL algorithm to wildlife image recognition can capture more detailed and accurate wildlife information and thus better help managers identify and monitor wildlife and protect the ecological environment and species diversity.This paper started with two aspects of public datasets and field data acquisition,analyzed the research status of deep learning,and introduced the research progress of the DL algorithm in wildlife image recognition.The paper focused on the present situation of regional convolutional neural networks(R-CNNs)and YOLO algorithms,so as to provide a theoretical basis for more efficient wildlife image recognition and offer new ideas for image recognition.

杨拂晓;费龙;闫泰辰

长春师范大学 地理科学学院,吉林 长春 130032吉林省林业勘察设计研究院,吉林 长春 130022

测绘与仪器

深度学习(DL)卷积神经网络(CNN)野生动物图像识别

deep learning(DL)convolutional neural network(CNN)wildlifeimage recognition

《北京测绘》 2024 (009)

1237-1242 / 6

吉林省科技发展计划(20230203001SF)

10.19580/j.cnki.1007-3000.2024.09.001

评论