FOT-E
| Reference number | |
| Coordinator | Chalmers Tekniska Högskola AB - Chalmers Tekniska Högskola Chalmers magasin |
| Funding from Vinnova | SEK 4 040 117 |
| Project duration | November 2019 - September 2023 |
| Status | Completed |
| Venture | Traffic safety and automated vehicles -FFI |
| End-of-project report | 2019-03095engelska.pdf (pdf, 2563 kB) |
Important results from the project
The objective of the project has been to extract data in existing, collected video footage, from previous data collection projects, as well as to classify behaviors and movements, in order to create an improved database for future research. The database must also be able to be used for the validation of upcoming, new safety functions regarding driver attention in traffic. Hos SAFER now has an updated FOT database where phone use and yawning are automatically noted. This has resulted in a significantly improved dataset, available for future research.
Expected long term effects
An updated FOT database is now available with auto-labeled features. This has resulted in an improved dataset, available for future research at SAFER. For automotive safety systems, naturalistic data are necessary for the analysis of driver behavior as well as testing of passive and active safety systems. For the eye tracker industry, the result can be used in automatic lower-level annotated functions, to develop algorithms for higher-level functions, such as sleepiness based on yawns and eye blinks.
Approach and implementation
The project´s objective was to improve SAFER´s existing database, by extracting functions from video data. This was done through an interactive process, where manually marked behavior in the videos was used as training data for newly developed image recognition algorithms. Among other things, telephone use and yawning were calculated. In the analysis, the project validated a number of these functions for the enriched data set.