Pedestrian traffic counter
Tracking people effectively depends on the settings. For instance, tracking people on busy streets round-the-clock is not the same as tracking people indoors. In most cases, you can’t install special counters or cover areas with numerous cameras to get ideal capturing angles and resolution. Therefore, the previous generations of counters were helpless outdoors. Today advanced deep neural network recognition modules can effectively detect people in a frame, recorded from any angle, at limited resolution and poor light condition.
BitRefine Heads is a universal modular video analytics platform. A user adds recognition modules to the processing pipeline according to the task. In the case of a pedestrian traffic counter, a typical pipeline consists of the following steps: Image acquisition, which is usually an IP-camera. Then the neural person detector comes into play. It effectively detects people by their shapes even on the harshest background. After all, people detected, information is passed to the tracking module, which adds an ID to the detected person and tracks his movements until he goes away from the frame. Keeping assigned IDs intact even when several people are moving close to each in a crowd involves several levels of filtering algorithms. The most sophisticated of them utilize the power of the neural network to distinguish persons from each other by the clothes they wear.
After BitRefine recognition pipeline acquires all the track of people, the next step is to draw one or several lines that will play the role of counters. As soon as the track crosses the line, the system records this event together with a timestamp, direction and optionally screenshot of the person. Each line can be named according to the user’s preferences and this name will be shown as a filter in the reports tool.
The reports section of the BitRefine platform provides quick visual access to collected data. User needs just to specify “line_cross” as an event of interest and the system will build a chart on a timeline with numbers of line crosses. To see statistics for certain areas, the user chooses corresponding cameras or counting line names and apply them as filters. Under the charts, BitRefine always forms a table with details of each registered line crossing event. By clicking on a thumbnail, the user can immediately see a screenshot of any particular event.
Being a distributed recognition platform BitRefine Heads can cover numerous locations and provide aggregated reports from one workstation. In addition to built-in reports, the BitRefine platform supports connection to 3rd party DBs and alarming functions that extends the scope of use for the recognition platform.
Related Solutions
We’ve got more video analytics to share. Let’s be friends. But you first...