摘要
为了建立快速、经济、准确、可行的预测肉牛体尺、体重性状指标的方法,试验利用Image 2图像识别技术和岭回归模型预测肉牛的体尺、体重,并与实测值进行比较,经过交叉验证该模型估测结果的准确性。结果表明:试验肉牛预测体尺与真实体尺的相关性>40%,其预测平均偏差小于0.04m。利用岭回归模型预测体重,通过3倍的交叉验证获得93%以上的准确率。说明可以利用图像识别技术与岭回归模型直接预测肉牛体尺、体重。
In order to establish a feasible,rapid,economical and accurate method to predict the body size and body weight of beef cattle,Image2 technology and ridge regress model were used to predict the body length and body weight of beef cattle in this test,the results were compared with the actual measured values,and the accuracy of the model’s estimation results was evaluated through cross-validation. The result showed that the correlation between predictive body size and true solid size was more than 40%,and the average prediction deviation was less than0. 03 m. The accuracy of prediction weight can reach more than 93% though 3 folds cross validation using the ridge regress model. It indicated that image recognition technology and ridge regression model can be used to predict beef cattle body size and weight directly.
作者
岳萌萌
舒涛
王嘉博
刘利
王鹏
赵晓川
许珊珊
王春薇
柴孟龙
孙芳
钟金城
YUE Mengmeng;SHU Tao;WANG Jiabo;LIU Li;WANG Peng;ZHAO Xiaochuan;XU Shanshan;WANG Chunwei;CHAI Menglong;SUN Fang;ZHONG Jincheng(Institute of Animal Husbandry,Heilongjiang Academy of Agricultural Sciences,Harbin 150086,China;Key Laboratory of Animal Genetic Resources Protection and Utilization,Ministry of Education,Qinghai-Tibet Plateau,Chengdu 610041,China;Shangzhi City Mayan Township People's Government,Shangzhi 150621,China;Heilongjiang Provincial Animal Husbandry Terminus,Herbin 150069,China;Heilongjiang Animal Disease Prevention and Control Center,Herbin 150069,China)
出处
《黑龙江畜牧兽医》
CAS
北大核心
2020年第22期41-43,49,共4页
Heilongjiang Animal Science And veterinary Medicine
基金
国家肉牛牦牛产业技术体系建设项目(CARS-37)
国家重点研发计划项目(2018YFD0501700)
黑龙江省农业科学院“农业科技创新跨越工程”专项(2019CX-16)
黑龙江省农业科学院院级项目(2018YYYF024)。
关键词
肉牛
Image
2
岭回归模型
体尺
体重
beef cattle
Image 2
ridge regression model
body size
body weight
