| Feeding behavior of dairy cows is one of the key behaviors of dairy cow activities,reflecting the health status and feeding situation of dairy cows.Traditional manual observations are only for individual pastures,while large-scale pastures only rely on higher labor costs.In recent years,with the continuous development of digital technology and artificial intelligence,wearable devices have been used in various fields.The use of this method in animal husbandry can reduce labor and capital costs,increase efficiency and reduce losses.Based on this,this article proposes a method based on activity data and deep learning to identify the feeding behavior of dairy cows.By collecting activity data to analyze,process and use deep learning and other algorithm models to identify,judge the feeding status of dairy cows and estimate the feeding amount of dairy cows,combined with air pressure sensor to verify the results,this article mainly completed the following work:Firstly,collecting and preprocessing the activity data of cows.Cow activity data refers to the comprehensive data when cows are moving in static and dynamic conditions,including various data such as resting and running,eating and ruminating.This paper uses a low-power collector to collect the cow activity data and preprocess it as the basic data set for the subsequent algorithm.Secondly,the pre-processed activity data is processed twice according to different algorithm models,and then the data after the second processing is extracted and trained with different algorithms(two-dimensional convolutional neural network,etc.)in order to adjust the model parameters to make the recognition results more accurate and then estimate the feed intake of individual cows.Finally,the different algorithm models are compared and the method with higher accuracy and practicality is selected,and then combined with the air pressure sensor,the relationship between the change of the air pressure value and the change of the cow’s head position is monitored to verify the recognition accuracy of the algorithm. |
