Open AccessRecognition method of dairy cow feeding behavior based on convolutional neural network
Author(s) -
Zhefu Chen,
Xiaodong Cheng,
Xi Wang,
Mingshu Han
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1693/1/012166
Subject(s) - computer science , sliding window protocol , convolutional neural network , artificial intelligence , production (economics) , animal husbandry , machine learning , window (computing) , geography , economics , macroeconomics , operating system , archaeology , agriculture
With the continuous development of the Internet of things and artificial intelligence, smart animal husbandry has gained popularity in recent years. It can efficiently solve the labor cost and capital cost, and reduce the loss of pasture and herders. This paper proposes a method based on convolutional neural network to identify the feeding behavior of dairy cows, by detecting the feeding behavior of dairy cows, judging their health status, estimating feed intake, and estimating milk production according to a proportional coefficient. First, this article need go to the ranch to collect activity data and manually calibrate the cows; second, use the sliding window method to upgrade the one-dimensional activity data; finally, feed the two-dimensional data into the CNN, continuously adjust the parameters and lengthen data width, the recognition accuracy rate can reach 89.5%, realizing the recognition of the feeding behavior of dairy cows. The results show that the CNN-based dairy cow feeding behavior recognition method designed and implemented in this paper can realize efficient automatic recognition and has high accuracy.