Utilização da visão computacional para análise do comportamento e bem-estar de vacas leiteiras em alojamentos de freestall.

Abstract

Among the several systems and techniques used in monitoring and controlling animal production processes the computational vision and image processing has been studied and applied in animal production as well as in product processing commercialization. This image analysis became an important tool when studying behavior and welfare of dairy cows. The experiment took place in a commercial dairy farm in the region of São Pedro, SP, where 6 minicolor cameras were installed within the free-stall housing. The objective was to develop a software for recognizing the free-stall confined cow’s behavior and evaluate their welfare pattern. Images were registered every 10 min and stored in the computer using a image capture device model IVC -120G, with digital input/output, and processing 30 frames/s. The software for the image capture and analysis was developed using Delphi 5. As a result, the software was useful to analyze an specific image in 4s, and in one continuous hour of use, it was possible to analyze up to 900 images. This improved the research work, and was concluded that the developed software allowed to automatically identifying the animal’s movement within the free stall, as well as its position and feeding act.