Q1: If we have object detection, why we need image segmentation? If the object detection solves your problem, it is what you need and it is simpler.
What if we want to detect the road which is curving. (well, actually we can expand the object detection to predict a path instead of a box.) How about image manipulation like background removal? Try to create a tree to analysis what an object compose of. (some semantic analysis)
Q2: These 2 article compares the different approaches in object detection: