A loose-housing barn in Northern Germany with thirty-six Holstein Friesian dairy cows was fully monitored by video cameras. The barn was partitioned into the areas for lying cubicles, moving area, licking stone, cow brush, feeding and water troughs. From automatically detected cow activity, the cows’ proportional usage of the resources associated with these areas was determined. The resulting time series were modelled as autoregressive integrated moving average processes, and fifteen days predictions of usage of resources were made that reached high accuracies. These predictions might be helpful to detect changes within the herd and to maintain the resources.

摄像机对位于德国北部的一间散装牲口棚和36头荷斯坦黑白花奶牛进行了全面监控。谷仓被划分为可放置小隔间，活动区域，舔石头，牛毛刷，饲料和水槽的区域。从自动检测到的奶牛活动中，可以确定奶牛与这些区域相关的资源的比例使用情况。将所得的时间序列建模为自回归综合移动平均过程，并进行了十五天的资源使用预测，这些预测达到了很高的准确性。这些预测可能有助于发现畜群内的变化并维护资源。