Developing a Simulated Online Model That Integrates GNSS, Accelerometer and Weather Data to Detect Parturition Events in Grazing Sheep: A Machine Learning Approach.

摘要: In the current study, a simulated online parturition detection model is developed and reported. Using machine learning (ML)-based approach, incorporates data from Global Navigation Satellite System (GNSS) tracking collars, accelerometer ear tags local weather data, with aim of detecting events in pasture-based sheep. The specific objectives were two-fold: (i) determine which sensor systems features provide most useful information for lambing detection; (ii) evaluate how these might be integrated using ML classification to alert event as it occurs. Two independent field trials conducted during 2017 2018 seasons New Zealand, each used training validation, respectively. Based on objective (i), four identified exerting greatest importance detection: mean distance peers (MDP), MDP compared flock (MDP.Mean), closest peer (CP) posture change (PC). features, final was able detect 27% 55% within ±3 h birth no prior false positives. If sensitivity manipulated such that earlier positives permissible, this increased 91% 82% depending requirement single alert, or two consecutive alerts occurring. To identify potential causes failure, three animals investigated further. Lambing appeared rely social isolation behaviour addition PC behaviour. results study support use ML-based grazing This first known application Application knowledge could have significant impacts ability remotely monitor commercial situations, logical extension remote monitoring animal welfare.