Data-Driven Collaborative Intelligent System for Automatic Activities Monitoring of Wild Animals

Activity profiling is key to understand individual behavior and group dynamics for a species. To date, individuals monitoring is directly performed by the ethologist, leading to several limitations in the quantity and quality of the results. In this work, we propose a data-driven collaborative system for automatic remote monitoring of wild animals, in a challenging environment, properly designed to satisfy the ethologist's needs. This smart system fuses sensors data to perform an intelligent behavior identification, allowing for automatic activity profiling. As a case study, we leverage a dataset collecting time-series acquired by tri-Axial accelerometer and GPS applied to 26 baboons for 35 days, to identify running, walking, sitting, standing and feeding activities. The results obtained in terms of prediction accuracy and decision-making process interpretability show that the system can overcome the hostile environment's challenges, proving to be an effective support to perform smart remote automatic profiling.