On Time

The train station in Lausanne is filled with cameras that anonymously track people walking through its corridors. Used to estimate the congestion in the station, the analysis of average pedestrian trajectories is crucial to the improvement of the life of thousands of daily commuters. To do this analysis, we divide the train station into small regions of around one square meter linked together if they are adjacent on the ground. Using more than 42 million pedestrian positions, we determine when a region is sufficiently crowded at a given time and record the propagation of the congestion on the network. It yields to the creation of thousands of disconnected sub-networks, here colorized on the image. If you have ever been to the Lausanne train station, your trajectory is probably contained in what you see!

On Time - 2014. Copyright Kirell Benzi







Aluminium & Plexiglas


Data source

Real-time pedestrian tracking


Thanks to Alexandre Alahi for installing thecamera network and collecting the originaldataset of more than 42 million points.

Ecole Polytechnique Fédérale de Lausanne

Additional resource

If you would like to know more about the scientific study behind this artwork, you can read our paper: Principal Patterns on Graphs: Discovering Coherent Structures in Datasets.