ARM-based behavior tracking and identification system for group-housed pigs

Background: A smart monitoring system is essential to improve the quality of pig farming. A real-time monitoring system provides growth, health and food information of pigs while the manual monitoring method is inefficient and produces stress on pigs, and the direct contact between human and pig body increases diseases. Methods: In this paper, an ARM-based embedded platform and image recognition algorithms are proposed to monitor the abnormality of pigs. The proposed approach provides complete information on in-house pigs throughout the day such as eating, drinking, and excretion behaviors. The system records in detail each pig's time to eat and drink, and the amount of food and water intake. Results: The experimental results show that the accuracy of the proposed method is about 85%, and the effect of the technique has a significant advantage over traditional behavior detection methods. Conclusion: Therefore, the ARM-based behavior recognition algorithm has certain reference significance for the fine group aquaculture industry. The proposed approach can be used for a central monitoring system.