Keynote sessions

Summary form only given. Start-ups offering computer vision solutions in the field of surveillance typically suffer under a short half-life - while Infrastructure operators regularly complain about high false-alarm rates and other limitations of practical usage. These phenomena are obviously related to the technological difficulties on the one hand, the reasonable sized gap between “applied research” and application on the other hand contributes as well. After giving a short overview of the ASFINAG Video-System in general, the major challenges for traffic surveillance related to “automated video analysis” are discussed. The list of challenges is in a very subjective manner compared to statements from the scientific community and the state-of-the-art. This analysis leads to a brief description of the gap between application of computer vision for infrastructure operators and “applied” research within this field. Some of the major points out of this gap will be highlighted, like application-oriented testing, input quality measure, the importance of operator-friendly false-alarm rates, set-up effort for calibration, … Finally, a few thoughts are presented, which might increase the chances for a fruitful transmission of scientific results into real world applications.