Análise de imagens para a caracterização das atividades de vacas leiteiras dentro do galpão de confinamento

Today's animal production need the use of information technology and automation in the animal's rearing environment in order to analyze their interference both in the production and animal welfare. The objective of this research was to compare the efficiency of two methodologies of image analysis for evaluating the presence of dairy cows in specific places in the confinement shed. The experiment was done in a commercial farm and the cows were monitored by six cameras. Two ways of analyzing the presence of the cows in the pre- selected places (bedding, drinker and feeder) were used: images analyzed as seen in the computer screen (T1), and automatically by the software developed for capturing the images of the cows in the freestall shed (T2). The attendance data were analyzed using the technique of the principal component analysis and to compare the two methods, it was applied the Student t - test with 95% reliability for the mean count of the presence of cows in each studied area. No significant difference between the methods was found, and both methods were efficient for registering the presence of cows in the activities.