摘要
鸡只生病或受到应激惊扰后,活动量会有显著性异常。通过对鸡只群体活动量的实时监测,实现鸡群生病预警等异常情况提示,这将极大降低家禽养殖企业的经济损失。随着信息技术的发展,计算机视觉由于其非侵入性,以及呈现丰富信息的能力,已成为家禽实时自动化监控系统的一个有效监测工具。通过YOLOv5和改进的Deep SORT算法对鸡群进行多目标检测与跟踪,提取鸡只外观特征和运动特征,将鸡群的活动情况量化,由此判断鸡群活动量是否异常,为鸡群的福利评估提供重要技术支撑。
Abnormal activity of chickens will be occurred when chickens get sick or are disturbed by stress. Through real-time monitoring of chickens, we could be reminded of the emergence of chicken disease at an early stage or other abnormal activity prompts, which will greatly reduce the economic losses of poultry breeding enterprises. With the current development in information technologies, computer vision has become a promising tool in the real-time automation of poultry monitoring systems due to its non-intrusive and non-invasive properties, and its ability to present a wide range of information. On the basis of chicken detection and tracking through YOLOv5 and improved Deep SORT algorithms, the appearance and motion features of chickens are extracted and the activity data of chickens are quantified. Finally, the alarm method is proposed to judge whether there is abnormal activity, which provides important technical support for the welfare assessment of chickens.
作者
周小芹
吕嘉
金宇
Zhou Xiaoqin;Lv Jia;Jin Yu(College of IoT Engineering,Hohai University,Changzhou 213022,China;China Mobile Group Henan Company Limited,Zhengzhou 450000,China)
出处
《科学技术创新》
2022年第28期57-60,共4页
Scientific and Technological Innovation
基金
常州市应用基础研究计划(中补助)《半放牧养殖家禽群体活动与异常行为的视觉监测及评价》(项目编号:CJ20190053)。
