摘要
目前,高原牧区养殖牦牛大部分还采用传统放牧方式,需要人工统计牦牛数量,将高原牦牛图像目标检测技术用于日常管理,可减轻牧民的负担。针对高原多变环境下的牦牛放牧场景,利用YOLOv5目标检测网络进行牦牛目标检测。为保证实验的有效性,笔者自行采集并标注了牦牛样本数据。实验结果显示,基于YOLOv5的高原牦牛目标检测方法在检测精度和速率上都取得良好的效果。
At present, most Yaks in plateau pastoral areas still use traditional grazing methods. The image target detection technology of plateau yaks is used for daily management and monitoring, which can reduce the burden for herdsmen. Aiming at the yak grazing scene in the changeable plateau environment, yak target detection is carried out by using YOLOv5 target detection network. In order to ensure the effectiveness of the experiment, yak sample data are collected and labeled. The experimental results show that the plateau yak target detection method based on YOLOv5 has achieved good results in detection accuracy and speed.
作者
曹红
龚赛君
CAO Hong;GONG Saijun(School of Information Science and Technology,Tibet University,Lhasa Tibet 850000,China)
出处
《信息与电脑》
2021年第18期84-87,共4页
Information & Computer
基金
西藏大学研究生“高水平人才培养计划”“基于神经网络的高原生物目标检测”(项目编号:00060701)。
