How AI and drones transform reindeer counting

Traditionally, reindeer counting has been a resource-intensive and time-consuming task, often carried out with helicopters, snowmobiles, or ATVs. These methods are costly and can be stressful for the animals. Now, new technology points toward a more efficient, precise, and sustainable solution: drones combined with artificial intelligence (AI).

The ARC (Aerial Reindeer Count) project, led by NIBIO in collaboration with NORCE and Biodrone on behalf of the Norwegian Agriculture Agency, aims to test how drones and AI can streamline reindeer counting.

Drones are already a useful tool in reindeer herding—they are small, easy to transport, and provide an overview of scattered herds. They are used to inspect fences, locate carcasses, and even move animals in challenging terrain. By replacing helicopters and snowmobiles with drones, both climate emissions and costs are reduced, while animal welfare is improved.

To develop an AI model that can automatically count reindeer, researchers have collected extensive image material from drones in various terrains and seasons. The images are then analyzed by an AI tool that identifies and counts the animals. Researcher Sigurd Løkse at NORCE is developing the AI model that counts the reindeer:

– We have collaborated with reindeer herders to collect images and videos of reindeer in different terrains and seasons. The images show reindeer with summer and winter coats, with and without antlers, and in various colors and patterns. These images have been used to train an AI model to recognize reindeer, explains Løkse.

Video: Automatic counting of reindeer. Source: NIBIO

Training the model is time-consuming, but the results so far are promising. All images of reindeer have been carefully manually classified so the model can recognize the many colors and sizes of the animals. The model now counts reindeer with about 90 percent accuracy in snow-covered terrain and open vegetation. Animals in dense vegetation and white reindeer on snow remain a challenge. Løkse and his colleagues are working to improve the model by adding more verified data and fine-tuning the algorithms.

The counting process will involve drones flying over large areas and taking high-resolution images of reindeer herds. To ensure the best possible image quality, the project has also developed a user manual for drone pilots, with recommendations on routes, flight altitudes, and camera angles.

– The goal is for the technology to become available to reindeer herders through a platform where they can upload their own images and receive automated counting data. In the long term, the solution could also be used for official reindeer counts, which would provide significant time and cost savings, says project leader Gabriela Wagner at NIBIO.

The Norwegian Agriculture Agency commissioned the researchers for this project. In a news release, they state that the technology could eventually be used in the management of wild reindeer and other wildlife.

The next step is to add thermal images for winter counting, which will be especially useful during the dark season when it is difficult to detect reindeer with regular cameras.

Video: Use of thermal camera for counting. Source: NIBIO

The ARC project demonstrates how modern technology can contribute to more efficient and sustainable reindeer herding. With drones and artificial intelligence, the industry gains a tool that improves animal welfare, reduces environmental impact, and provides better decision-making support for both reindeer herders and management authorities.