Um modelo de visualização e rastreio de comportamento de animais durante o pastoreio

Abstract

Monitoring grazing animals is essential to increase productivity, promote animal welfare, and support sustainable livestock practices. However, there is still a gap in the availability of accessible tools and replicable methodologies for behavioral analysis in extensive grazing environments. This dissertation aimed to develop and evaluate a low-cost tracking methodology using both adapted devices and real-world datasets. The research was conducted in two complementary scenarios: (i) a pilot experiment using a smartphone adapted to collect data from a calf in pasture and (ii) the analysis of five datasets provided by the company Inovafarm, including the monitoring of up to 198 animals over periods of up to ten months. Based on these data, the computational tool AgroTrack was developed to process, filter, and visualize information through heatmaps, trajectories, and dynamic videos. The results demonstrated the feasibility of using affordable devices, the scalability of the solution for large herds, and the importance of continuous monitoring to identify movement patterns and preferential grazing areas. Overall, the integration of accessible devices with computational analysis techniques can expand access to precision livestock farming, benefiting both small-scale and large-scale producers.