Journal of Applied Poultry Research (Dec 2022)
Automated video sensing – real world correlations
Abstract
SUMMARY: The main objective in rearing animals is to maximize the efficiencies during production while maintaining their welfare. There is also now significant interest from consumers to understand the conditions under which animals are grown. Assessing welfare is a challenge due to the extensive number of factors that affect welfare; and there is currently no agreement on quantitative measures to describe suitable performance in animal management that are not based on growth performance. Most farm managers use visual observations of behavior as one of the significant features for assessing the state of the animals. Advances in imaging technology allow for easier and more accurate ways to compute and evaluate different indicators by using visual cues.This work looks at the potential to use region feature comparisons by using features derived from video data to aid in conducting these assessments in the environment. This approach supports a more holistic approach to management that incorporates the traditional measures to optimize the production system.Manual analysis of videos can be labor intensive and automated techniques could improve efficiencies and provide significant benefits. In this work, correlations between features generated from the video analysis and physical observations were determined to be significant. The use of this technology aids in the measurement of parameters that could be instrumental in defining behavior and could provide quantitative descriptors of management outcomes. However, more research needs to be conducted associating the outputs of new image processing algorithms to metabolic and physiological changes in the birds. Quantifiable measures such as these could them be used to more consistently describe conditions under which birds are being raised allowing researchers and producers to make informed decisions in designing integrated systems for production management.
