Where is my Deer?-Wildlife Tracking And Counting via Edge Computing And Deep Learning

Abstract

Reliable, informative and up-to-date information regarding the location, mobility and behavioral patterns of animals enhances our ability to preserve biodiversity, manage invasive species and conduct research. The basis of which is an accurate count of the animals present in a specified region. In literature, previous studies have presented automated animal counting methods, usually relying on using single images. Thus, accuracy is challengeable due to several factors, including wildlife movement, light fluctuations, overlapping, occlusions and re-counting of the same animal reappearing in other images. In this paper, we present a novel approach of identification, introduction to tracking pipeline, and counting wildlife accurately. Having applied the techniques of deep convolutional neural network (CNN), edge computing, and online tracking in a field trial to determine the population density of deer in a given area. Our approach produced accurate and actionable results, thus there is viable commercial potential.