Desire lines normally refer to paths created by humans as a result of excessive use, as that path is the most efficient form of transportation. So if I mount my Pi Camera looking at an intersection, how could I map the desire lines for that intersection using my RasPi and camera? I am more thinking about tracking where people come from to get to a specific point.
IIUC you’re trying to establish patterns of life using a camera to track individuals. A Pi I think will be hard pushed to process the imagery to recognise separate individuals. You might be better using pressure pad sensors and multiple Pi spread around the area. You could then fuse the pressure pad readings to establish individual tracks. That is a non trivial mathematical problem esp when you’re tracking multiple individuals. Tool wise Python may not be the best solution. I’d look at using Wolfram Mathematics and Google the likes of track before detect for the fusion aspects. Good luck, you’ll need it!