Elaboration of Basic Methods for Automatic Analysis of Cows’ Gait

Two different methods for automatic registration and analysis were used to produce data for comparison and analysis of lame and healthy animalsʹ gait in Estonia. A walk over mat with two quazi-piezoelectric sensors was elaborated and tested in co­operation with University of Helsinki. Preliminary analysis indicates that lameness can be seen as asymmetric gait and thus the quazi-piezoelectric walk-over mat is a promising tool for automatic leg problem detection. A video-system was introduced to record walking pattern of cows in co-operation with Catholic University of Leuven. For video recordings three cameras were used to obtain top, side and leg views with StreamPix software video-signal capture. Possibilities of image based separation of dairy cows with real time vision system and preliminary settlement of this was developed. A model-based motion scoring system is proposed for derivation of image parameters needed for lameness detection. About 600 cows once a week were investigated in a large dairy farm during four monthsʹ period. Dairy cowsʹ gait pattern was recorded with the aid of quazi-piezoelectric walk-over mat and video-system. Preliminary lameness scoring was performed in the cowshed visually by two experts. These scoring results were later specified by expert commission on the basis of video-recordings. Lameness scores (according to Sprecher et al) were assigned as follows: 1-6,012 cases, 2-1,181 cases, 3-522 cases, 4-105 cases and 5-37 cases from total 10,653 cases. The database of cowsʹ identification numbers, lameness scores and disordered legs description was created, that allows synchronization of walk-over mat signals data and video files.