Research On Body Size Measurement And Body Condition Of Ujumqin Sheep Based On Machine Vision

Ujumqin sheep is an excellent local breed and a precious resource of meat sheep breeds in my country.The body size index is a key index reflecting the physical growth performance,appearance and health level of sheep.Using traditional methods to perform artificial body measurement and appearance identification is not only time-consuming and labor-intensive,but also increases the incidence of zoonotic diseases.In view of the above problems,this study carried out correlation analysis,path regression analysis,and error comparison between real values and machine vision values for 51 rams and 172 ewes in Ujumqin breeding farm.Based on the analysis and reference to the body shape scoring methods of dairy cows,beef cattle,and dairy goats,as well as the Ujumqin sheep national standard for the classification of body shape and appearance,the guideline for the construction of the Ujumqin sheep body shape record is initially proposed.The main conclusions are as follows:（1）In order to quickly obtain the data of the oblique length,body height,chest depth and cross height of the Ujumqin sheep,this study uses the method of machine vision recognition and applies the Mask-R-CNN convolutional neural network to process the image data.,through comparative analysis with manual measurement data,it was found that the average relative errors of body oblique length,body height,chest depth and cross height of 33 Ujimqin sheep were 3.89%,4.13%,5.39%,and 3.78%,respectively..The relative error rate was less than 5%except for chest depth greater than 5%.On the whole,compared with manual measurement,the data fitting based on machine vision has smaller aggregation errors and higher accuracy.（2）Through the SPSS analysis method,it was found that the correlation between the weight and the body size of the Ujimqin ram was extremely significant and the ranking results were:chest circumference>chest depth>body oblique length>cross height>body height>body width（p<0.01）.The extremely significant relationship between weight and body size of Ujimqin ewes was:cross section height>body height>body oblique length>bust circumference（p<0.01）.Therefore,the key body size parameters of concern for the production performance measurement of rams and ewes are different,and different selection and breeding programs should be developed to guide production practices efficiently and continuously.（3）Through R language and SPSS software analysis method,the difference,significance and correlation between body size and body weight of Ujumqin sheep were analyzed.The optimal linear regression model of the weight of Ujumqin rams and ewes is constructed,that is,the optimal regression model of the weight of rams is:Y=-85.4905+0.7677X₁+0.8532X₆+0.6967X₇;The optimal regression model of weight is:Y=-103.20689+0.43044X₁+0.95643X₄+0.59232X₆+0.26107X₇+2.09554X₈.This model provides theoretical basis for the selection of Ujumqin sheep.（4）In view of the lack of a complete and unified recording guide for the body shape record of mutton sheep,this thesis takes Ujimqin as the research object and combines the body shape recording standards of dairy cows,beef cattle and ICAR dairy goats to initially construct the body shape of Ujumqin sheep.Guidelines are documented and a list of recommended standard traits is given for reference in production practice.