Research And Application Of Pig Breeding System Based On Image Recognition

In recent years,with the frequent outbreaks of swine flu and classical swine fever,people have become very sensitive to pork safety.Pig farming is one of the important pillar industries to implement the goal of getting rid of poverty and getting rich and promote rural rejuvenation,but there are still limitations of traditional farming methods and lack of support and integration of the latest information technology.How to ensure the safety of pig source and feeding process has become an urgent problem for pig breeding enterprises and consumers.In view of the above problems,this study established a pig breeding system based on image recognition to provide information services for pig breeding,thus reducing the cost of pig breeding.In this paper,through consulting literature and field investigation,we analyzed various behaviors in the production and life of farmed pigs,separated the classic pig behaviors through statistics,and defined them: stand,lie down and take food.In view of the above three behaviors,a large number of image data are obtained by manual shooting and network collection,and the unavailable image data are deleted after screening,and finally the data set is determined.The collected and sorted pig data set is expanded by data enhancement methods such as Flipping,Rotation,Noise Injection,Color Space,etc.,and labeled by Label Img software,which is divided into three labels: stand,lie down and take food.Then,the objects in the labeling box are labeled by behavior categories and saved in the file marked with pictures.Secondly,the YOLO V7 model of target detection in deep learning is used to identify three kinds of behaviors of pigs:standing,lying down and eating.By analyzing the data results,we can observe the growth of pigs and pave the way for the establishment of pig breeding system platform.Finally,by using the Internet of Things technology,a pig breeding system platform is established,which can be divided into PC and mobile terminals,and can be used in farms,enterprises and animal husbandry bureaus.The PC end of the system can store user information,basic management information and basic data of pigs,and realize various functions according to user login module,background management module,breeding distribution profile and other modules;The APP end of the system is divided into four sections: business management,real-time monitoring,expert service and aquaculture consultation,which provide convenience for users.This study makes the breeding process of enterprises more standardized,the feeding data more systematic,and can ensure the breeding quality of each pig,and bring healthy pork to the majority of consumers.At the same time,it also provides reference and reference significance for the development of pig breeding and the establishment of other animal husbandry,and hopes to improve the development level of pig breeding in China.