Validation of a commercial system for the continuous and automated monitoring of dairy cows activity

In order to improve animal welfare and enhance the comfort of dairy cows, the application of information technology (IT) within the intensive livestock farming takes a key role in a proper routine management. This study aims to compare localisation and activity data provided by the CowView system, an automatic indoor localisation system for dairy cattle, with those obtained by a manual labelling procedure, twice within an observation period of minimum 25 hours per dataset. Data from five selected dairy cows were represented by behaviours performed in relation to the occupied zones, and were classified in two categories: activity and localisation. The identified activities performed by the dairy cows were standing, walking (both considered as being in the alley), resting (being in the cubicle) and feeding (being at the feeder). Indeed, the zone considered in the analysis were alley, in bed and feeding zone. Data automatically and manually classified (used as a reference) were compared. Among all the behaviours detected by the automated software, the most reliable results are those related to the activity of feeding (accuracy higher than 95%). The results showed that the CowView automatic monitoring system is able to identify activity zone classification (ALLEY, THROUGH, CUBICLES) with higher reliability compared to the specific activities performed by dairy cows. The results obtained support the CowView system as an innovative and effective solution for an easier management of dairy cows.