Measuring People Flows

Measuring comprehensive and accurate quantitative data of people’s real motion behavior is of utmost importance for both analyzing people flows and developing people flow simulation models. arsenal research investigates which measuring systems are suited for specific environments and needs, and applies them in close accordance with legal privacy regulations.

Directional People Counts

The number of people passing through cross sections such as doorways, elevator doors and escalators is a valuable source of information. While there are a number of solutions based on infrared, laser or video systems that are already commercially available, it eventually remains unclear how accurate the counts delivered by these devices are for given environments. This is especially true when people densities are high, doorways are broad and counts are needed several times a day within relatively short counting time intervals. arsenal research evaluates the statistical error characteristics of counting devices by capturing ground truth data with manual annotations of video footage. When observing a systematic bias compared to ground truth data, the performance of the counting devices can be improved. Based on the research new counting methods are developed or existing systems improved.

Click on the following image for an example evaluation of the error statistics of our award-winning approach for vision-based counting in a crowded subway scenario.

Zoom

The approach has been developed in cooperation with Joanneum Research, Austria. The correction factor determined by this evaluation improved the accuracy of the vision-based counting algorithm up to 98 %. Such evaluations can be applied to any counting technology.

Spatio-temporal People Motion

arsenal research uses a variety of approaches to obtain rich spatio-temporal motion datasets. This includes the capturing of GPS data for outdoor applications, Cell-ID data from mobile phones, laser range data and cutting-edge solutions for vision-based people tracking and crowd analysis.

The following images illustrate an example of strategic planning for video camera installations in a train station (left) and the output of a vision-based people tracker (right).

Zoom