In order to evaluate pork parameters and pork grade objectively,this paper adopted BP artificial neural network in neural network toolbox of MATLAB,used hog's carcass image features and living hog's image features to establish BP neural network model of BP.Then trained 60 samples of hog's carcass image features and 80 samples living hog's image features,got the neural network model of BP.And adopted different BP neural networks to imply the transmission function of layer and implied the quantity of one layer of neurons to the artificial result and trained the influence of errors of the goal to compare correctly.Then compared simulation result and the artificial result.The result indicates BP artificial neural network model can assess pork every index and grade discern.Appraise the rate of accuracy of the pork grade and is up to 98% under the pork trunk vision characteristic index,appraise the rate of accuracy of the pork grade and is up to 80% in pig's vision characteristic parameter of living body.It proves that pork trunk vision characteristic can represent pork quality than living body pig vision characteristic parameter accord with the objective reality.The result indicates that at the same time the BP artificial neural network in MATLAB neural network toolbox can be applied to the grade evaluation of the pig.