奶牛体况评分目标检测数据集  

A dataset of target detection for dairy cow body condition scores

作　　者：黄小平 豆子豪 黄飞 郑寰宇 侯现坤 王晨洋 冯涛 饶元[2] HUANG Xiaoping;DOU Zihao;HUANG Fei;ZHENG Huanyu;HOU Xiankun;WANG Chenyang;FENG Tao;RAO Yuan(National Engineering Research Center for Agro-Ecological Big Data Analysis&Application,School of Internet,Anhui University,Hefei 230039,P.R.China;Key Laboratory of Agricultural Sensors,Ministry of Agriculture and Rural Affairs,College of Information and Computer Science,Anhui Agricultural University,Hefei 230036,P.R.China)

机构地区：[1]安徽大学,互联网学院,农业生态大数据分析与应用技术国家地方联合工程研究中心,合肥230039 [2]安徽农业大学,信息与人工智能学院,农业农村部农业传感器重点实验室,合肥230036

出　　处：《中国科学数据(中英文网络版)》2025年第1期24-35,共12页China Scientific Data

基　　金：安徽省科技厅自然科学基金项目(2308085MC103);安徽省教育厅自然科学重点项目(KJ2021A0024)。

摘　　要：奶牛体况评分是评价奶牛产能与体态健康的重要指标。目前,随着现代化牧场的发展,智能检测技术已被应用于奶牛精准养殖中。本研究旨在创建一个奶牛体况评分目标检测数据集,以支持农业领域中奶牛健康监测的自动化。本数据集由体况评分3.25、3.5、3.75、4.0和4.25的5类图像组成,且每张图像都标注了奶牛的体况评分区域。体况评分是衡量奶牛健康状况的重要指标,对于精确而高效的评估具有重要意义。数据集的构建来自规模化养殖农场的奶牛图像,每张图像都经过专业兽医团队的体况评分,评分结果被用于生成标注信息。标注信息包括奶牛体况评分的边界框坐标和相应的评分标签。为了提高数据集的多样性和鲁棒性,在数据集中引入了一系列挑战性因素,例如不同光照条件、奶牛姿态的变化以及背景杂乱的情况。这有助于确保训练的目标检测模型在真实场景中具有良好的泛化性能。通过开发这一奶牛体况评分目标检测数据集,为农业领域的奶牛健康监测提供了一个有力的工具,有望推动相关技术的发展,提高奶牛管理的效率和精度。此外,本数据集还为研究人员提供了一个有价值的资源,用于深入研究奶牛体况与健康之间的关联,为未来的农业科技创新提供支持。Body condition score is an important indicator for evaluating the productivity and health of dairy cows.With the development of modern farms,intelligent detection technologies are now applied to the precision breeding of dairy cows.This study aims to create a dataset of target detection for dairy cow body condition scores.dairy cow body condition score target detection dataset to support the automation of dairy cow health monitoring in the agricultural field.The dataset consists of 5 types of images with body condition scores of 3.25,3.5,3.75,4.0,and 4.25 for multiple cow images,each labeled with a body condition score area.Body condition score is an important measure of cow health,which is of great significance to accurate and efficient evaluation.The images were collected from large-scale farms,with each cow scored by a team of professional veterinarians.The scores were then used to generate annotation information,which includes the bounding box coordinates of the body condition score area and the corresponding scoring label.To enhance the diversity and robustness of the dataset,we introduced a number of challenging factors into the dataset,such as different lighting conditions,changes in cow posture,and background clutter.This helps to ensure that the trained object detection model exhibits good generalization performance in real-world scenarios.The development of this dataset can provide a powerful tool for cow health monitoring in the agricultural sector,which is expected to advancements in related technologies,improving the efficiency and accuracy of cow management.In addition,the dataset serves as a valuable resource for researchers to delve deeper into the connection between body condition and health in dairy cows,supporting future innovations in agricultural technology.

关 键 词：奶牛体况评分 目标检测 农业领域 健康监测 

分 类 号：S823[农业科学—畜牧学] TP391.41[农业科学—畜牧兽医]