I don’t remember any street tracking libraries, but can you provide more details on what exactly you want to do? Do you have a top view on the streets or is the camera located on a vehicle, etc.? Given a top view image, do you want to build a graph containing streets, or do you want to extract such information from a video sequence?
In the first case, you can try using edge detection algorithms and see if you can extract the corresponding edges from the image. Or maybe you can try using the color or the intensity of the cut to leave only the streets in the image if they are different from those around them.
In the second case, you have image / video metadata. What are the coordinates of the camera, in which direction does it look at each frame / image?