Browsing by Subject "kuntoluokitus"

The increasing stock sizes and generalization of automatic milking systems have decreased the contacts between animals and humans and also decreased the amount of working time that can be used to observe cow’s condition. Especially in large stocks body condition scoring is very laborious. Body condition scoring can provide useful information of the cow’s energy reserves and help the farmer to optimize the cow’s feeding and thus milk production. Various studies based on automatic image analysis present that body condition scoring can be automated. There is also a commercial application for automatic body condition score. The aim of this study was to develop an automatic machine vision-based body condition scoring (BCS) system in feeding station for dairy cows, determine which variables can affect to the reliability of image analysis and how well the rear area of the cow’s back and cow’s height correlate to the cow’s BCS. 17 holstein- and 4 ayshire-cows were filmed and measured in Natural Resources Institute Finland’s research farm. A video camera was installed above to feeding stations entrance to film the cow’s rear back area and an ultrasonic-sensor was installed inside the feeding station to measure cow’s height. An algorithm was developed by using a Matlab-platform for the image analysis to detect and extract the cow’s rear back from the image and to measure the area of the cow’s rear back. The image analysis managed to detect and extract reliably only brown-white ayshire-cows from the image. A problem occurred with the black-white holstein-cows. The hue and contrast between the holstein’s black color and the background of the image was not sufficient. The lack of hue and contrast was mainly caused because of the dim light conditions of the barn. Because of the failure in image acquisition and image analysis only 7 observations of the cow’s back area were sufficient. In the present studies the number of observations was not enough to model cow’s body condition score reliably. However the method used in this study confirms the challenges in image analysis and results observed in previous studies. The use of cow’s rear back’s area and height in BCS is plausible, but the method needs more observations and better placement of the filming and measuring devices.