Análise da movimentação de frangos de corte em diferentes temperaturas baseada em vídeos

Abstract

Extreme temperatures, which induce prostration and death of broiler chicken, cause not only losses to the vast Brazilian agriculture, but also suffering to the animals. Automated monitoring of their movement can, however, help avoid losses and assist in decision-making by breeders. In this work, we aim to propose and evaluate an approach for the automatic analysis and behaviour identification of videos of the chickens at different temperatures in a controlled environment. Through videos of the birds, following simple restrictions, reports are generated, then fed to a visualization software to analyze the difference in chicken movement at different temperatures in a clear and comprehensible way. The technique used to extract data from the videos was detection and tracking through video processing software using color-based detection of spots marked on the birds, and the CSRT (Channel and Spatial Reliability of Discriminative Correlation Filter) algorithm for tracking. To create information and generate a report from the data, basic principles of statistics and data analysis were used, such as the concepts of median, highest value, etc., and graph analysis. The results show that there are statistically significant differences in movement according to temperature, which is expected and demonstrates the effectiveness of the system in correctly evaluating the movement of the birds.