[campus icon] Accesskey [ h ] University of Paderborn - Home
EN english
Die Universität der Informationsgesellschaft


New Dataset for Hazmat Labels
Datum: 2018/07/08
Link: https://osf.io/b5dap/

GET Lab has developed a novel high-quality hazmat label test database. Hazmat labels are signs that indicate certain dangers (e.g. "explosive") and their detection is of interest in the field of rescue robotics. The new dataset is intended to stimulate research in this area. It has been used in our publication "Seeing Signs of Danger: Attention-Accelerated Hazmat Label Detection" by Mohamed Mahmoud, Jan Tünnermann, and Bärbel Mertsching, which will be presented at the IEEE International Symposium on Safety, Security, and Rescue Robotics 2018 in Philadelphia, PA, USA. To benefit the community, the dataset is now available via the Open Science Framework: https://osf.io/b5dap.

Three exemplary test images from the dataset. A Simple target “Dangerous when wet” on relatively homogeneous woodchip wallpaper background B “Flammable liquid” label on OSB wood wall. C “Radioactive” label on sunny outdoor brick wall