Model-based calving monitor using real time image analysis

In this study a model-based calving monitoring was developed to identify the standing and lying behaviour of cows during their pre-calving period. For this purpose video surveillance of 8 cows for 6 discrete hours during the 24 hours prior to their calving were analyzed. Using the model-based calving monitor, 7 different shape variables through time were obtained for each cow. These were: x and y coordinate of the centre point of the cow; walking trajectory; distance walked; orientation of the main axis; body width/length ratio; hip length and back area. These variables were then used to classify standing and lying behaviour. On average 85% of the standing and lying behaviour of the 8 cows during the last 6 discrete hours before calving could be correctly classified.