Video analysis (Parking Management):

The demonstrator showcases a scenario in which moving objects in a car park, primarily vehicles, are visually located and tracked using the infrastructure-mounted lane cameras. Our system processes video streams from multiple sources in a car park. Once motion is identified in a video stream, an object detection module analyzes the respective image regions to check for false positives. Then, the detected object is further tracked by a compressed-domain algorithm, whose results are translated into global coordinates by a georeferencing module. The controller orchestrates this interaction and finally, the visualization component displays the results in a sensor-specific detailed 2D and global aggregated abstract 3D visualization.