Abstract: The livestock breeding industry is developing towards large-scale and intensive development pattern. The intelligent recognition of livestock behavior can provide important data for livestock disease prevention, rational feeding and breeding, etc., and improve the efficiency of livestock breeding management. As an efficient image information processing technology, computer vision technology can provide a non-contact and automatic way of monitoring livestock behaviors. This article reviewed the research progress of computer vision technology in animal behavior recognition from three aspects: livestock behavior recognition visual system, livestock target extraction and individual recognition, and introduction of livestock daily and abnormal behavior. Finally, the challenges and future development trends of computer vision technology in livestock behavior recognition were summarized, in order to promote the intelligent development of livestock breeding.