Conclusions
Optical flow patterns can provide an easy way of analysing the
behaviour of large groups of animals fromvideo records. The optical
flow measures described here (mean, variance, skew and kurtosis)
did not track individual animals, but characterized the overall
movement of whole flocks of chickens. Nevertheless, the optical
flow patterns detected at group level, especially the skew and
kurtosis of movement, were correlated with behaviour and welfareat the individual level, such as whether a high percentage of
chickens were walking abnormally. Optical flow thus provides an
information-rich link between individual and group and potentially
has wide application to the study of flocks, herds and other animal
groups as an alternative to other more intrusive or invasive
methods.