Methods for Automated Remote Sensing and Counting of Animals

Abstract

Remote sensing and counting of animals is important for both livestock farming and wildlife population management. Traditional surveying methods for manual onsite counting are hard to accomplish, time-consuming, expensive and could be even dangerous. In contrast, remote detection and counting is safe, cost-effective and could be easily and frequently repeated, providing prompt information about animals’ population size and location. This paper reviews some of the most promising potential methods for automated remote detection and counting of animals, ranging from simple computer vision operations like spatial filtering, edge detection, binarization and image subtraction, to more complex machine learning and AI approaches for object detection and recognition. Many scientists worldwide are working on automation of farming and livestock management, and their experimental results really prove the great potential of computer vision and artificial intelligence for achieving that.