Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Abstract. Monitoring the individual body weight is of great importance for pig health management and pig production systems. The objective of this paper was to achieve the estimation of live body weight without contact and stress on animals. The proposed approach was based on 3D data provided by the Microsoft Kinect V2. The camera was placed over the pen in which the pigs lived freely, rather than as a single-row channel or limit fence. Top-view images of Landrace gilts and Yorkshire shoats in group-housed were processed to extract body features. The SVR models using leave one out cross-validation developed with data including gilts and shoats presented an R² ranging from 0.95 to 0.98. While using the merged data sets, the R² was ranged from 0.97 to 0.98. The RMSE and MAE values of ultimately selected VA-Q-SVR model were 5.5969, 4.299%. These results revealed the approach has the potential for 3D computer vision to estimate body weight in the pig breeding industry.