Dynamic time warping for classifying lameness in cows

Abstract

Lameness in dairy cows is one of the most significant welfare and productivity issue. This work is mainly concerned with an automated detecting system for classifying lameness in dairy cows. In the proposed system, Dynamic Time Warping (DTW) is used to measure the similarity between two-time series. The first time series is the behavioral time periods of the cow used as the templates, which was collected while the cow was sound. The second time series is the behavioral time periods of the cow on each day used for testing. This process results in accumulated distance that is compared with a threshold value for classifying lameness. In the case of studies, three cows were used in experiments. The classified results show that the proposed algorithm can correctly classify lame and non-lame cows.