Research On Measuring Method Of Cattle Body Size Based On Deep Learning

In recent years,with the development of animal husbandry,the cattle industry that occupies an important position in animal husbandry has also shown a rapid development trend.The body size parameters of cattle mainly include body height,body length,and oblique length,which not only can directly reflect the growth and development status of cattle,but also are important indicators for cattle breeding and meat quality evaluation in fine breeding.The traditional cattle body size measurement mostly uses manual contact measurement.The stress response of cattle is large,which results in difficult measurement,large errors and low efficiency.To solve this problem,this paper proposes a non-contact measurement method of bovine body size parameters based on deep learning.This paper proposes a method of measuring bovine body size parameters based on deep learning.It uses the deep learning model to detect six types of feature frames of cattle,and realizes the extraction of local feature contours of bovines,the extraction of body feature points,the correction of cattle standing posture,and the calculation of body size.The main work of this article is as follows: the image data set of the characteristic parts of the cattle body size is produced,which contains more than 5000 images with the parts of the cattle,the cattle head,the trunk,the cattle koji,joints and hoofs.The Faster RCNN algorithm is used to detect the body feature parts in the image;the Canny edge detection algorithm is used to extract the edge contour from the local image of the feature parts.Use the least squares method to fit the quadratic curve of the edge detection points output by Canny edge detection to obtain the smooth contour of the cow’s characteristic parts;on the obtained edge contour line,use the method of calculating the maximum curvature to obtain the measurement point of the cattle body size;The four hoof coordinate information is used to realize the correction of the standing posture of the cow;the parameters of the body height,body length and body oblique length are calculated by combining the calibrationparameter set.Through experimental verification,this method can effectively measure the bovine body size parameters,and the relative errors of the measured bovine body size parameter values are all less than 4.5%.The deep learning cattle body size measurement system studied in this paper can realize the non-contact measurement of cattle body size parameters during feeding.In this paper,based on deep learning algorithm to study the measurement of non-contact cattle body size,and provide theoretical basis and experimental support for the method.The application of artificial intelligence to the field of animal husbandry expands its development in this field,and provides a new way for the measurement of livestock body size parameters.