摘要
针对牲畜牛健康检测中的牛脸快速检测问题,提出一种基于改进单镜头多盒检测器(single shot multibox detector,SSD)模型的牛脸检测方法。以轻量级网络MobileNet v2作为模型的基础网络,较使用VGG-16的SSD模型,速度有一定提高;使用K-均值聚类对模型默认框的高宽比进行优化,使模型更加具有针对性。实验结果表明,和其它方法相比,该方法在精确率、召回率、FPS等指标上较优。
Aiming at the problem of rapid detection of cow face in the health detection of livestock cattle,a method based on improved SSD(single shot multibox detector)model was proposed.To increase the speed of model,the lightweight network MobileNet v2 was selected as the basenet of model instead of VGG-16.K-Means clustering was used to optimize the aspect ratio of the model default box to make the model more targeted.Results of experiments indicate that the proposed method is superior to other methods in terms of accuracy,recall rate and FPS.
作者
苟先太
黄巍
刘琪芬
GOU Xian-tai;HUANG Wei;LIU Qi-fen(School of Electrical Engineering,Southwest Jiaotong University,Chengdu 611756,China)
出处
《计算机工程与设计》
北大核心
2020年第3期833-837,共5页
Computer Engineering and Design
基金
四川省重大科技专项基金项目(18ZDZX0162)
四川省重点研发基金项目(2017GZ0159)。
