Digital Video Tracking Stats

I am looking to create an application that can use a live video feed and extract statistical information about human movement. The example of a public library works best to explain the requirements. We need to track the number of entries and exits into and out of the library. We need to understand the movement of people within the library. Where do they go and how long do they spend in each isle. It will need to track multiple people simultaneously. We need to overlay the video image with coordinates or a grid so we can cross reference movement with isle information. The video will be in a fixed position overhead, looking down like a floorplan.