Statistical tool for the estimation of body weight of lactating cows from 2D photographs

Abstract

Estimating the body weights of animals is an arduous task in livestock management. However, it is a vital tool in developing suitable breeding programmes,calculating the nutritional requirements, determining pharmaceutical doses of drugs and assessing feed conversion efficiency and carcass quality. In this study, we made an attempt to predict the bodyweights of lactating cows using images from various angles. Seventeen lactating cows from the University Livestock Farm and Fodder Research Station, Mannuthy, were selected randomly. Different body measurements of the animals were obtained manually and photographs of the lateral, rear, and dorsal views were taken using the demo version of the CAD-KAS measure-pictures 1.0 software. The surface area was estimated using the body length along the back on the midline from the withers to the croup and the maximum stomach width from the dorsal view photographs. A stepwise multiple linear regression model was used to predict the body weight of the animals. A regression equation fitted using the surface area and perpendicular length between the Tail Head above and the plane in level with the ventral line of the abdomen below yielded an R2 of 90.5 percent. This study suggests that models built using measurements retrieved from digital image analysis techniques can be used as a promising tool for predicting the body weight of animals. Keywords: Cattle, body weight, digital image analysis, multiple linear regression