Application of Machine Learning and Analytics for Cattle Behaviour Classification within an Internet of Things Deployment

Lameness is a big problem in the dairy industry, farmers are not yet able to adequately solve it because of the high initial setup costs, vendor incompatibility and complex equipment in currently available solutions, and as a result, this work presents a hybrid model and an end-to-end Internet of Things (IoT) application that leverages machine learning and data analytics techniques to predict lameness in dairy cattle. As part of a real world trial in Waterford, Ireland, 150 cows were each fitted with a long range pedometer. The mobility data from sensors attached to the front leg (left leg for 50% of the cows and right leg for the other 50%) of each cow is aggregated to formtime series of behavioral activities (Step count, lying time and swaps per hour). These are analyzed in the cloud and alerts of predicted lame animals are sent to the farmer’s mobile device using push notifications. The application and model automaticallymeasure and can gather data continuously such that cows can bemonitored daily. This means there is no need for herding the cows as this would bias the results because cows are stoic in nature. Furthermore the clustering technique employed proposes a new approach of having a different model for subsets of animals with similar activity levels as opposed to a one size fits all approach. It also ensures that the custom models dynamically adjust as weather and farm condition change as the application is extended to other farms. The initial results indicate that the application can predict lameness 3 days before it can be visually seen by the farmer with an overall accuracy of 87%. This means that the animal can either be isolated or treated (usually by administering antibiotics) immediately to avoid any further effects of lameness. The application designed in this study is based on a fog-to-cloud architecture. In this architecture, some of the cloud services and applications are run closer to the physical IoT devices at the network edge. The application also implements a microservices based design approach. The solution can therefore be decoupled as a single service which can be accessed via an Application Programming Interface (API) either by the farmer seeking such a service or an agri-tech service provider who wants to provide such a service to his exiting customers. This also aids data preprocessing and aggregating between the fog node and the cloud. The result of this show an overall data reduction from 10.1MB to 1.62MB exchanged between the fog node and cloud node daily. This is the first time such an approach is implemented for lameness detection and generally for welfare monitoring for dairy cattle