ESTIMATING LIVE WEIGHT OF PHILIPPINE DAIRY BUFFALOES (Bubalus bubalis) USING DIGITAL IMAGE ANALYSIS

Abstract

Two regression models were developed to predict the live weight of dairy buffaloes (Bubalus bubalis) using digital image analysis. The first model relates the live weight with measurements of body length (BL), wither height (WH), fore girth depth (FD), hip height (HH), rear depth (RD), and diagonal length (DL). The second model relates the live weight to the number of pixels present in a segmented digital image  of  the  animal.  Body  measurements and pixel counting were done by processing the digital images of the animal using ImageJ. Stepwise regression analysis revealed that BL, WH, and FD significantly affected the changes in live weight of dairy buffaloes at P < 0.05. Further analysis showed that dimension-based regression model (R2= 0.94 and RMSE = 22.12) performed better than pixel-based model (R2= 0.87 and RMSE = 23.22) in estimating live weight of dairy buffaloes.