This probably is not help, but more of a question for understanding:
If you have you movement threshold sensitivity set to the point that a car will not trigger it, how can a smaller target like a person trigger it?
Having said that, the fact that a person will not trigger at all may be because the person, at a distance, is simply not a large enough target. Movement in a video frame has to be de-sensitized to some degree, otherwise tree/bush movement, changing sunlight patterns from rapidly moving clouds, etc will produce many false triggers. I know software detection is a lot better than when I used it in early 2000s but my guess is the software still looks for some percentage of the image to vary for a trigger and the targets are too small to hit that percentage. Zoom in and see what happens when a person walks across the bridge.