cv-people-detector

A project for detecting people in video data.

Background

Have you ever wondered how people move around your city? Do you want to measure the foot traffic in front of your store? How about how often certain areas of a public park is used? That is the question we proposed in the beginning of this project. We were tasked to tally the patrons and the areas that were most commonly used in Weequahic park. With this data, park managers can address requests for additional facilities and better manage resources.

In the first phase of this project, we chose to use an urban intelligence platform called Placemeter. It uses video information to quantify pedestrian, cyclist and vehicle movement. As a first step, we chose to measure the traffic of the jungle gym right below the fieldhouse.

In the second phase of this project, we will be developing software for detecting people from video feeds. This could be used to count how many people use walking paths, sports fields, or tracks.

Progress from phase I

We used Placemeter to moniter the traffic at the Weequahic park jungle gym for a month (August). We collected the resulting data into a Google fusion table. Just for fun we compared the number of people walking through the jungle gym area to the daily temperature. Our results below seem to suggest fewer people go to the park when it's hot out. The interative version of this plot can be found here from this Google fusion tables spreadsheet.