There are (some, or many) shopping malls that have installed cell-phone trackers of various sorts, the idea being to know how people move around inside the mall, and even inside individual stores. They use what-ever info they can get from the phone (which I'm thinking is wifi and bluetooth MAC address) to do this. They don't know who owns the phone, but they get pretty good info about how often that person (or who-ever is carrying the phone) attends the mall, which stores, which isles or departments, for how long, etc. For what-ever good that does them in terms of knowing what's popular, traffic, etc.
Naturally this is dependent on how people have their wifi and bluetooth setup, but I've read some reports that over 50% of smartphones have wifi roaming turned on - so they can seemlessly (mindlessly) connect to their home or office networks whenever in range.
You might also want to look up a concept known as "bluetooth marketing".
The problem with trailcam's is that they take too long to power up from "sleep" mode. During sleep mode, most of the camera is powered down, with only the IR sensor being powered all the time. When the IR sensor detects movement, it powers up the rest of the camera to take the picture. When the camera is used normally (in a forest or bush) and you're trying to take pictures of deer or other slow moving animals, the trail cam works fine. However, when you want to detect people that are walking quickly through the sensor's field of view, quite frequently you won't get a good image of the person because by the time the camera is ready to take the picture, the person has (or has almost) walked right through the field of view.
I've thought that a really good urban security-cam would have these features:
1) ability to mount IR sensors adjacent to the camera, so that you give the camera more advanced warning of approaching objects so it can turn itself on and be ready to take the picture. How these separately-mounted sensors communicate with the camera (wires? wireless?) is an issue. Knowing how to hack into (connect) separate sensors into one (or more) commonly available (under $75) trail-cams is one potential hack-a-day project).
2) ability to give off a short burst of sound just before the picture is taken would be VERY useful. A beep, a pre-recorded animal sound, etc. Something that would make a person walking in front of a trail cam turn his head in the direction of the sound as a picture is taken. Having a good picture of a person's face is key to following up with police. How to integrate something that can make the desired sound into a trail cam as part of it's motion-detection and picture-taking sequence could be a separate hack-a-day project. In the most ideal case, a stand-alone mp3 player (something the size of a USB thumb-drive) that can playback an audio file to a speaker using a trigger signal.
3) build a remote IR LED array that can be positioned more effectively to illuminate an approaching object. The array would be powered on by a signal hacked from inside a conventional trail-cam. The trail cam's own IR array could be deactivated if it was determined that the remotely-mounted array was more effective.
4) as stated in the first post, detect an electronic signature of an approaching cell phone, and either log the information (the project being simply to log cell phones as they come and go) or to activate a trail-cam to capture a photo of the person carrying the phone. In the most sophisticated hack, a battery-powered trail cam would have wifi/bluetooth MAC address or imei detection capability and incorporate that text into any pictures it takes.