基于图像深度学习的牛个体的识别与统计  

Cattle identification and statistics based on images deep learning

作　　者：何钦 李根 严永煜 于文静[1] 王嵘 He Qin;Li Gen;Yan Yongyu;Yu Wenjing;Wang Rong(School of Information Science and Engineering,East China University of Science and Technology,Shanghai,200237)

机构地区：[1]华东理工大学信息科学与工程学院,上海200237

出　　处：《电子测试》2021年第21期68-71,28,共5页Electronic Test

基　　金：2020年度上海市大学生创新创业训练计划(S20075)资助项目。

摘　　要：为进一步协助畜牧养殖业,对传统物理方法及先进的生物检测方法进行了多方面的调研,并从图像深度学习角度出发,基于Google TensorFlow框架及OpenVINO^(TM)工具套件对牛个体的识别与统计进行了研究。利用神经网络对数据集进行训练,并使用SSD目标检测算法对目标对象进行识别,希望为相关行业的监管提供图像处理角度的解决方案并对之后的研究提供一定的参考价值。In order to assist animal husbandry,the traditional physical methods and advanced biological detection methods were investigated in many aspects.From the perspective of image deep learning,the recognition and statistics of cattle were studied based on Google TensorFlow deep learning framework and Intel OpenVINO^(TM) tool kits.Neural network is used to train the data set,and SSD target detection algorithm is used to identify the target object,hoping to provide a solution from the perspective of image processing for the supervision of related industries,and provide some reference for future research.

关 键 词：牛个体识别 谷歌TensorFlow SSD目标检测算法 英特尔OpenVINO工具套件 

分 类 号：S823[农业科学—畜牧学] TP18[农业科学—畜牧兽医] TP391.41[自动化与计算机技术—控制理论与控制工程]