文章基本信息

标题：Image Technology based Cow Identification System Using Deep Learning
本地全文：下载
作者：Thi Thi Zin ; Cho Nilar Phyo ; Pyke Tin 等
期刊名称：Lecture Notes in Engineering and Computer Science
印刷版ISSN：2078-0958
电子版ISSN：2078-0966
出版年度：2018
卷号：2233&2234
页码：320-323
出版社：Newswood and International Association of Engineers
摘要：Today worldwide trending in precision dairy farming is becoming more focus on an individual cow welfare and health rather than group management by using modern technologies including image processing techniques. In such cases, individual cow identification is one of the fundamental ingredients for the success of modern dairy farming. Thus, in this paper we shall explore and examine how image processing technologies can be utilized in analyzing and identifying individual cows along with deep learning techniques. This system is mainly focus on the identification of individual cow based on the black and white pattern of the cow’s body. In our system, firstly we detect the cow’s body which have been placed on the Rotary Milking Parlour by using the inter-frame differencing and horizontal histogram based approach. Then, we crop the cow’s body region by using the predefined distance value. Finally, the cropped images are used as input data for training the deep convolutional neural network for the identification of individual cow’s pattern. The experiments are performed on the self-collected cow video dataset which have been taken at the large-scale ranch in Oita Prefecture, Japan. According the experimental result, our system got the accuracy of 86.8% for automatic cropping of cow’s body region and 97.01% for cow’s pattern identification. The result shows that our system can automatically recognize each individual cow’s pattern very well.
关键词：cow identification; deep learning; edge detection; horizontal histogram