Estimation of Body Surface Temperature Change of Swine Using Thermography by Machine Learning Approach

Abstract

It is a big loss if overlooks swine estrus in breeding work in swine farm. Whether breeding swine are in estrus is a large work load for workers since it is needed to observe carefully many swine based on their experience in breeding work in swine farm. Previous research has shown that whether breeding swine are in estrus can be checked by measuring vulvar surface temperature change using thermography as a quantitative method. However, it was reported that temperature is easy to be changed by infl uence of external environment such as ambient temperature of swine. In this paper, we extract 5 external environment factors; gluteal surface temperature of swine using thermography, the temperature inside and outside swine farm. And we estimate vulvar temperature eliminated the infl uence of their factors. We performed 3 regression analysis by machine learning method, and evaluated the combination their factors and method for estimating. As a result, we could estimate vulvar temperature whose error is less, than 1.4 °C using gluteal temperature in swine, the temperature inside and around swine farm as external environment factors as explanatory variables. And we found this method can estimate temperature change by estrus correctly.