Video recodings of traffic are an extremely rich source of information. Since early 2000, Transport & Roads tightly co-operate with the Mathematical Imaging group (Centre for Mathematical Sciences website) at LTH developing tools and methods for interpreting video images into meaningful data that has value for traffic researchers and practitioners. The development goes in the following areas:

• Detection and tracking of road users
• Detection of events of interests, e.g. traffic conflicts
• Accuracy of the position and speed measurements taken from video
• Use of different camera sensors (RGB, thermal) and combination of sensors (e.g. stereo vision)
• Intelligent quantification of behaviour patterns - spatial, temporal and related to interactions

This co-operation resulted in several PhD theses defended on the topic and a number ofopen-source software tools released.

To comply with GDPR (EUR-Lex website) and Swedish Camera surveillance law (The Swedish Parliament's website, in Swedish), our group relies primarely on low-resolution RGB or thermal videos which do not allow recognition of individual vehicles or persons. As such, no personal data is collected.

We consult and co-operate with a number of commercial initiatives developing video analysis products for intelligent traffic analysis.