AI preserves biodiversity and prevents wildlife accidents

In collaboration with the Swedish University of Agricultural Sciences, the Swedish Transport Administration and the Norwegian Institute for Nature Research, Sweco has developed a tool that uses AI to distinguish different animal species in images from wildlife cameras.

Wildlife cameras are used in research to document animals and their behavior, for example in wildlife population surveys and wildlife crossing monitoring. This is needed to plan safe passages for animals over and under roads and railways to avoid collisions with vehicles, and to avoid disturbing animals in their habitat.

Since wildlife cameras are activated by movement, a huge amount of footage is also created of precipitation, falling leaves, vehicles and people. As a result, a lot of time has been spent manually going through all the images and sorting the material. In addition, there was no secure filing of the imagery. Capture is the solution to this.

Using algorithms, Capture identifies relevant images and sorts them so that researchers only need to verify the selection. Capture also ensures privacy protection by anonymising images of people and vehicles.

About the collaboration

Capture has been developed in a collaboration between Sweco, the Swedish University of Agricultural Sciences, the Swedish Transport Administration and the Norwegian Institute for Nature Research. SLU works with wildlife research and provides the Swedish Transport Administration with information that enables planning for safe passage of animals over and under roads and railways, wildlife traps, outcrossings and improved fencing. Twenty-five projects in Sweden and Norway use Capture in their research.

Thanks to its modular structure, Capture can be used in more areas than wildlife. With other algorithms, it would be possible to identify invasive species from drone images, for example . –  Kaisa Ekqvist, digitalisation expert and project manager at Sweco