MIDAS - anonyMIsing DAta collection for traffic Safety
| Reference number | |
| Coordinator | RISE Research Institutes of Sweden AB - RISE Mobilitet och System, Göteborg |
| Funding from Vinnova | SEK 7 094 592 |
| Project duration | April 2020 - March 2023 |
| Status | Completed |
| Venture | Electronics, software and communication - FFI |
| End-of-project report | 2019-05873eng.pdf (pdf, 6781 kB) |
Purpose and goal
This project has focused on developing anonymization methods that enable industry and the public sector to collect high-resolution data in a socially sustainable way by using innovative anonymization algorithms that do not degrade data resolution. These methods aim to replace the removed personal information with computer-generated information so that images look naturalistic and non-personal information, such as gaze direction and facial expressions, is preserved.
Expected results and effects
The project has developed anonymization algorithms for faces and number plates. It has been possible to demonstrate the functions on both image and video data. For face anonymization, it is possible to replace faces in video clips with great fidelity and the ability to keep the same face over time in the sequence. It is also possible to retain some non-personal information such as facial expressions and gaze direction.
Planned approach and implementation
All parties were involved in driving the project forward and the development. Berge contributed a PhD student who was a leader in the technical development and publication of articles. Halmstad University has offered supervision for the PhD student. RISE has acted as coordinator. Viscando has acted as problem owner and provided data.