Image recognition for wildlife cameras
| Reference number | |
| Coordinator | Trafikverket - Trafikverket, BORLÄNGE |
| Funding from Vinnova | SEK 454 400 |
| Project duration | May 2020 - November 2021 |
| Status | Completed |
| Venture | AI - Competence, ability and application |
| Call | Start your AI-journey! For public organizations |
Important results from the project
The aim of the project is to create a user-oriented solution for management, analysis and interpretation of images from fauna-related projects at the Swedish Transport Administration and other authorities. The project developed a IT-platform "Capture" that combines the manual management of metadata with automatic image recognition through AI / ML and automated depersonalization of images of people or vehicles. Capture has been developed and tested in consultation with ongoing camera projects at the Swedish Transport Administration and SLU. Further development is in progress.
Expected long term effects
Capture streamlines the handling, analysis and categorization of images from camera studies of fauna measures and other fauna projects. It offers a centralized archiving of depersonalized image material and thus a basis for evaluating changes in the environment and infrastructure. The modular system enables new algorithms for image recognition of other than wildlife to be easily incorporated and thereby expand the area of use. The reference group including the EPA, county boards and hunter associations, propose Capture to become a central service at SLU.
Approach and implementation
Capture was developed and tested on SLU´s IT platform through collaboration between SLU, Sweco AB and the Norwegian Institute for Natural Research (NINA). The image recognition module uses algorithms developed by NINA that can be re-trained using manually validated images from the platform. The result is an archive of cleared, depersonalized and categorized images with associated metadata that can be exported to other programs or to statistical analyzes to, for example, calculate species-specific visit frequencies at monitored fauna passages.