Livestock detection in African rangelands: Potential of high-resolution remote sensing data

Abstract

Livestock production is vital in eradicating poverty, malnutrition, and in attainment of the Sustainable Development Goals (SDG) in developing regions such as Africa. The livestock sector of Africa contributes 10%–44% of the gross domestic product and more than 70% of the greenhouse gas emissions of the continent. With the anticipated increase in demand for livestock products, the need to mitigate climate change, and lack of accurate livestock census data, innovative remote sensing technologies and methods for livestock census become crucial for the livestock sector. In this paper, we present a review of current technological advancements in remote sensing and detection algorithms in livestock censuses, identifying weaknesses in sensors and detection methods, and highlighting issues that currently limit adoption of these technologies in African countries. We observed that the last four years (2019–2022) accounted for 69% of all livestock detection studies. This surge was driven by development of Unmanned Aerial Vehicles, which offer high resolution images and flexibility for detection. In addition, the use of automated detection methods are fast, efficient and accurate. However, the surrounding background of different livestock species, herd size and spatial resolution of the datasets affects detection accuracy. We suggest the need for publicly accessible aerial labelled livestock databases covering the various livestock breeds in Africa to develop customized detection models for the heterogeneous landscapes in the rangelands. Efficient detection methods are vital for monitoring livestock population trends and environmental impacts of grazing practises.