Research and Application of Dairy Cows Body Condition Score Based on Attention Mechanism

Abstract

In order to solve the problem of low accuracy in body condition scoring of dairy cows with a small number of training samples, an attention mechanism based cow body condition scoring method was proposed in this paper. The posterior and lateral posterior image information of dairy cows were collected to score the body condition of dairy cows. In the traditional VGG16 convolution neural network is introduced into CBAM (Cnvolutional Block Attention Module) Attention mechanism, improve the convolutional neural network model of cow body of information channels and spatial focusing ability, make the feature extraction model can notice to cows of key characteristics, the characteristics of the compression is not important, then can improve the cow body condition feature extraction ability, applied to the cow body condition scoring method in the study. The deep learning model was trained, and the experimental results showed that VGG16+CBAM algorithm could improve the accuracy of cow body condition scoring. Among them, this method has the best effect on the body condition evaluation of cows from the positive rear Angle, and the accuracy can reach 95.60%.