Classification Of Pig Attitude Behavior Based On Sparse Coding Pyramid Model

China is a long-established agricultural country.As part of the development of agricultural animal husbandry,the pig industry plays a positive and important role in protecting people’s material life,so more and more people are keen to pay attention to this cause.With the continuous development and improvement of computer science and technology,the level of computer intelligence has gradually increased,and the application fields have been extensively extended to the work of the pig industry,especially related technologies of computer vision.In the traditional pig industry,it is found that the abnormal behavior of pigs is mainly labor-based,which is time-consuming and labor-intensive and subject to subjective influence.Therefore,the modern pig industry uses intelligent monitoring technology instead of manual observation of labor.However,the RFID(radio frequency identification technology)used in actual production will cause harm to the physiology and psychology of pigs.Therefore,artificial intelligence pigs using computer vision technology were born.It is a machine learning method with sparse coding algorithm and A technical method combined with a pyramid model proposed by a professor in the United States to assist in tracking analysis is used to classify various poses of pigs.According to the order of image acquisition,image processing and gesture classification,the main work is as follows:First,collect color images of pigs from the Jingdong Financial JDD Grand Tour official website through web crawler technology.The collected images are then pre-processed,sparsely encoded and spatially pyramided to obtain a complete,smooth image of the pig’s pose binary.Then,this paper analyzes the types of pigs to be studied and selects five common pig poses.Including lying down,standing sideways,looking up,standing sideways and standing upright,erected and bowed.By using the convolutional neural network classification method to classify the pig’s action posture,and the comparative analysis finally obtained the average classification accuracy above 95%,indicating that this method is still very accurate in judging the pig’s action behavior,there is a certain Value meaning.This paper proposes to classify the pig’s posture by intelligent new computer vision technology to judge the actual behavior of the pig.This method has an important contribution to the modernization of the pig industry.The observation and recognition of pigs by computer greatly reduces the labor force and the errors that may be caused by labor.When the pigs have abnormal action behaviors,they can accurately and timely judge and process them accordingly,which is highly efficient and avoids many unnecessary losses.