Desenvolvimento de ferramenta não invasiva para dar subsídios na avaliação do bem-estar animal em aviários

Abstract

The poultry production in Brazil, for domestic consumption and export, has been increased, at a time when the concern for animal welfare is growing. The demand for information about the production process and trade of animal products involving ethical and commercial issues. The study of animal behavior has an important role in production, especially for the development of production systems that have minimum standards of welfare. The visual observation of behavior is time-consuming, subjective, and susceptible to error. In this way, the objective of this work is the use of cameras and image processing techniques to calculate the movements and count the density of the broilers in the production system. The images were acquired from an aviary from the western region of Paraná between 12/30/2016 to 2/14/2017. An infrared camera brand Hikvision model IR Fixed Dome was used. A total of 34,417 poultries from the COBB lineage were housed. At the end of the production process, 33,208 poultries with a total weight of 92,680 kg were collected, generating an average of 2.7 kg per poultry. The welfare evaluation process was developed based on two items: (1) the amount of kilos of poultry per m2, where Brazilian legislation indicates a maximum value of 42 Kg/m2, in the best scenario. For this evaluation, the Haar-like and LBP classifiers were used, both were trained with 200 and 400 positive images, and their validations were performed visually. The classifiers analyzed 1,920 images, in this process were identified how many poultries were inside a frame of an m2. (2) the ability of movement, where Brazilian legislation and EFSA and Global G.A.P. Indicate that there should be space for the animal to move and rest. For this item, tracing techniques were used one based on Optical Flow and another one called KCF. The tracking techniques were tested through the evaluation of 32 videos, four videos per week, the evaluation recorded the movement of the poultries. Through the proposed approaches it can be concluded that it was possible to generate information to subsidize the evaluation of animal welfare in aviaries through the applied techniques.