Automatic identification of marked pigs in a pen using image pattern recognition

• Published in: Computers and electronics in agriculture Vol. 93; pp. 111 - 120
• Main Authors: Kashiha, Mohammadamin, Bahr, Claudia, Ott, Sanne, Moons, Christel P.H, Niewold, Theo A, Ödberg, F.O, Berckmans, Daniel
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01-04-2013; Elsevier
• Subjects: Agronomy. Soil science and plant productions; Algorithms; Animal productions; Automated image processing; Automation; Biological and medical sciences; Ellipse fitting; Fittings; Fundamental and applied biological sciences. Psychology; image analysis; Image processing; Labelling; Pattern recognition; Pig identification; piglets; Pigs; Resting behaviour in pigs; Terrestrial animal productions; Vertebrates; Automated image processing; Pattern recognition; Pig identification; Ellipse fitting; Resting behaviour in pigs; Image recognition; Farming animal; Image processing; Fitting; Pig; Automatic processing; Vertebrata; Mammalia; Artiodactyla; Ungulata; Automatic
• ISSN: 0168-1699; 1872-7107
• DOI: 10.1016/j.compag.2013.01.013

The purpose of this work was to investigate feasibility of an automated method to identify marked pigs in a pen in experimental conditions and for behaviour-related research by using image processing. This study comprised measurements on four groups of piglets, with 10 piglets per group in a pen. On average, piglets had a weight of 27±4.4kg at the start of experiments and 40kg±6.5 at the end. For the purpose of individual identification, basic patterns were painted on the back of the pigs. Each pen was monitored by a top-view CCD camera. Ellipse fitting algorithms were employed to localise pigs. Consequently, individual pigs could be identified by their respective paint pattern using pattern recognition techniques. Taking visual labelling of videos by an experienced ethologist as the gold standard, pigs could be identified with an average accuracy of 88.7%. It was also shown that behaviours such as resting can be monitored using the presented technique.