Abstract: In order to solve the problem that the sheep are easily frightened due to the close flying distance when the UAV is tracking the sheep, the author proposed the method of judging the status of sheep by tracking sheep with UAV, that is, the UAV adopted YOLOv5-DeepSort（you only look once-DeepSort）algorithm to locate and track the sheep. This method combined MSK（mean shift-K-means） algorithm of Mean shift and K-means clustering algorithms to judge whether sheep were frightened. The results showed that YOLOv5-DeepSort algorithm could be used to accurately locate and track sheep, and MSK algorithm could accurately judge whether the sheep were frightened. The results indicated that the judgment method of sheep status could be used to determine the basis of the adjustment of the UAV tracking distance, so that the UAV would not scare the sheep while tracking.