SIS Software

SIS Software

Published November 2019.

The project partner SIS has been commercially active in the field of people flow analysis at major events for many years. Here, for example, smartphone apps are used to record movement data of visitors to the event and evaluate it in real time in order to provide rescue forces, for example, with information on crowded areas of the site as quickly as possible. This project will now explore the extent to which crowd flow predictions are possible when only short-term training data – rather than data from multiple previous iterations of the event – is available.

SIS Software GmbH is currently the market leader with its technology for recording the behavior of flows of people using smartphones. The next natural step in this topic is the prediction of future crowd movements within a short-term period. However, the current state of the art here either requires extensive reference data to create a valid model of the movement flows or is simply not precise enough to withstand commercial requirements. In the project, based on the dataset, a more novel modeling of people flows will be explored, where simple behavioral patterns (e.g., the behavior after passing through a gate) are learned based on actual training data and these patterns are fused into a larger model. The goal is to create a model that allows valid short-term predictions of crowd behavior after only a few hours of “observing” the event

For the project, two datasets from the Rock am Ring events from 2018 and 2019 (each with about 20,000 attendees over 3 days), as well as a dataset from a large event in Zurich (about 17,000 attendees) will be provided. The data set from Zurich can also be reused for follow-up projects in the future.

The project results are to be published in a high-quality publication. In addition, the results are to be directly incorporated into commercial products of the project partner. This will directly lead to an increase in safety at large events, as critical situations (e.g. blockages at gates and passageways) could be detected earlier than is currently the case.

Project period

01.11.2019 – 30.04.2020


Dipl.-Inf. Gernot Bahle (, DFKI

Dr. Tobias Franke (, SIS Software GmbH