摘要
针对常见动物识别的问题,提出了一种基于YOLO的识别分类系统。通过搭载linux系统的树莓派完成实施检测运行的识别程序。利用YOLO算法进行目标物体检测,其单个神经网络能够在一帧图像中直接预测边界边框(bounding box)和分类概率(classified probability),最后根据输入图像来输出动物物种的识别结果。其识别准确率针对鸟类可达94.61%,狗类为90.60%,牛类为79.03%。
To solve the problem of species identification and population identification of small animals,a recognition and classification system based on YOLO(You Only Look Once)is proposed in this paper.The recognition program is implemented on a Raspberry Pi with Linux system,and YOLO is used as the detection framework.The single neural network can directly predict the bounding box and the classified probability in a frame image,and output the animal recognized according to the input image.The accuracy rate of identification is 94.61%for birds,90.60%for dogs and 79.03%for cattle.
作者
周文萱
胡龙桃
张敏
方宇涛
李欣钰
Zhou Wenxuan;Hu Longtao;Zhang Min;Fang Yutao;Li Xinyu(School of Optoelectronic Information and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)
出处
《计算机时代》
2019年第3期22-25,共4页
Computer Era
关键词
物种图像识别
动物识别
YOLO
神经网络
树莓派
species image recognition
animal identification
YOLO
neural network
Raspberry Pi
