Farmers and managers of large natural areas need to know how many animals there are and where they are. For instance, in the Netherlands, farmers receive subsidies to protect meadow birds and promote biodiversity, which involves managing the grazing of their livestock. This means that grazing might be restricted on certain grasslands, and only a limited number of animals can graze there. The RVO - the agency that manages these subsidies - is responsible for monitoring and enforcing these rules. However, inspectors currently check only 5% of these subsidised areas, and they can only do this once or twice a year.
We can use remote sensing technology to monitor livestock. With artificial intelligence (AI) we can count and identify different animal species using data from satellites, aerial photographs, and drones. With more and better satellite data, higher resolution aerial photos, and advancements in drone technology, we can improve the efficiency of animal monitoring. This makes it easier and quicker to keep track of livestock, reducing the need for time-consuming field inspections.
To show how useful these technologies can be, we conducted a pilot study using three types of remote sensing:
- Satellite Data (50 cm per pixel, frequently available worldwide)
- Aerial Photographs (7.5 cm per pixel, taken once a year)
- Drone Imagery (~1 cm per pixel, available on request but expensive)
We used different AI techniques to automatically detect animals. We found that cattle can be detected in satellite data, but the results are sometimes inaccurate. For more precise detection and to identify specific species, we needed aerial photographs. Drone images allowed us to distinguish individual cows. These techniques are not only useful in the Netherlands but can be applied worldwide in large natural reserves or extensive farming areas with thousands of animals.
