Research on deep learning-based automatic monitoring technology of chicken cough in caged chicken house

Respiratory diseases are one of the diseases that are prone to occur in today's intensive livestock and poultry farming in a high-density farming environment, which has a huge impact on farm production. The traditional way of raising chickens is by manually entering the chicken house to patrol the pen to listen to the sound, and sometimes it is even necessary to enter the house to listen to the coughing sound at night after the chickens have rested. The manual inspection method is time-consuming and labour-intensive, and the time for actually listening to the sound is very limited, which makes it not easy to detect the problem in a timely manner in the pre-morbidity period. With the development of sensor technology, Internet of Things technology and intelligent algorithms, it is possible to make automatic monitoring of chicken cough. At present, there are mature products on the market for monitoring the cough of pigs, but there are no mature products for monitoring the cough of chickens, and there is an urgent need to develop a cough monitoring system for caged chicken sheds. In this paper, we selected the cough of chickens in caged chicken sheds as the research objective, and designed a system that can timely and accurately monitor and identify the coughing sound of sick chickens in caged chicken sheds by means of capsule information network, bidirectional recurrent information network, deep learning of the sound recognition system and other technologies, which can replace the traditional method of manually inspecting the fence to listen to the sound, and realise the monitoring of coughing sound in a highly efficient way with less time and effort.