Initial Feedback from Autonomous Counting and Placemeter

Post date: Nov 20, 2015 4:21:58 PM

After a month of work in Venice, we have been able to collaborate with our partners in the Venice Project Center, including Piero Toffolo and Professor Carrera, to install 9 of the 10 cameras that we came over with. This experience has taught us a great deal about what it takes to properly format, setup, and make the D-Link 933L wireless cameras accesible by the software run by Placemeter, the company we partnered with in New York before leaving the country.

The way that Placemeter measures pedestrian flow is by creating measurement points, which are lines superimposed over the camera streams. Each camera view has the possibility of having multiple measurement points, so cameras that overlook larger squares can have lines that count each entrance or street. Each of these lines produces an individual data stream, including distinguishing bidirectional traffic flow.

As we have had the opportunity to install these cameras in a variety of locations throughout the city, we have noticed trends in the performance of the cameras and the data returned from Placemeter's software. Based on our observations, the cameras with strong wifi signal tend to be much more accurate than cameras with weaker wifi. This is an issue that could be solved by having the Placemeter software operate with existing wired networks, therefore removing the variable of wifi. An ideal scenario would be obtaining access to existing networks such as the ones currently operated by the public transportation company (ACTV) or the municipality of the City of Venice.

Attached below is a graph showing the average pedestrian flow in (blue) and the average pedestrian flow out (orange) from one of our cameras placed within the city. This average is taken from a week of data.