Pig Posture Recognition Based On Acceleration Sensor

With the development of modern pig farming in the direction of scale and intensification,pig activities and health conditions are of great concern.The traditional manual monitoring of pig health will consume a lot of time and energy,and cannot get accurate monitoring results,and is also easy to cause cross-infection of human and animal diseases,which is no longer applicable to large-scale pig farming.In recent years,with the development of computer technology and the continuous reduction of sensor costs,modern information technology has been widely used in the field of pig health monitoring.The posture behavior of pigs must appear during their growth process,and the research on pig posture found that pig posture can not only analyze the behavior of pigs such as estrus,delivery,and lactation but also provide a reference basis for pig health judgment.Therefore,the use of pig posture behavior to determine the health status of pigs has become a hot spot for research in the field of pig health monitoring.For this reason,this paper proposes to reduce the human resource requirements and improve the level of automated monitoring of farms,etc.for pig posture recognition,and then builds a system based on acceleration sensors for pig posture recognition of its health.To this end,we first propose a pig posture recognition scheme and design the hardware and software parts of the system;then we use the Support Vector Machine(SVM)algorithm to classify and recognize pig posture;finally,we propose a pig health assessment method based on posture recognition to assist the breeder to better understand the pig health condition.The main research contents of this paper can be summarized as follows.(1)Pig posture recognition scheme design.A pig posture recognition system based on an acceleration sensor is designed to meet the demand for pig posture recognition in farms,and then the hardware and software parts of the system are designed.The hardware part mainly includes the sensor module,a processor module,transmitter module,and power module,and the selection of each module device and circuit schematic design are made according to the system requirements;the software part of the system includes the acquisition node,transmission node,database,and front-end interface design,and the code writing,database construction,and front-end interface development are completed to realize the acquisition,transmission,analysis,and display of pig posture data.(2)According to the actual observation on the farm,the pig posture categories were divided into four categories: fast walking,lying down,stationery,and slow walking.The three-way acceleration coordinate system of pigs was established using the principle of acceleration sensor detection.The pig posture classification was verified using manual analysis of experimental data.(3)A pig posture classification model based on the SVM algorithm was constructed,and a pig posture recognition system was designed by applying the SVM classification model,and the pig posture recognition was carried out with fattening pigs as the validation model.The experimental results show that the classification model has an accuracy rate of over 84% for pig posture recognition,which achieves the purpose of pig posture recognition,and the posture classification study lays the foundation for the subsequent pig health assessment.(4)Given the close and complex relationship between pig health and its growth stage and feeding conditions,the law between pig health and its posture was analyzed,and the application conditions of pig posture recognition for health assessment were derived.And taking fattening pigs as an example,based on the pig posture recognition,we applied the pig posture recognition to realize the assessment of its health condition based on the characteristics of the statistical pig posture duration.The above work shows that the acceleration sensor-based pig posture recognition system can collect posture data from a long distance and recognize pig posture with high accuracy,and also evaluate pig health,so that breeders can understand the health condition of pigs in time,and can detect the factors threatening pig health early,which in turn can help prevent the spread of virulent infectious diseases in the pig herd,which is of great significance for disease prevention,reducing losses in pig farms,improving pig welfare and promoting the development of smart pig farming.