Not just sight, but bone structure in this case.
You can look at a Panda's bone structure and see that it is not in the bear family. There are lots of little structural details that help out.
To give an idea of how advanced the science is --- you can often tell the race of a human (same species) from bones by little things like sinus cavtity space, femur head relative thickness compared to the knee joint, and lots of pretty obscure details.
Same general concept here.
That's a bit misleading. We have a huge baseline of humans for comparative analysis. When you start dealing with fossils of extinct animals, you have to start making assumptions. You start saying things like: OK, this tooth is similar to a pig, therefore X. OK, this femur is similar to an ostrich, therefore Y.
It becomes a great deal of educated guesswork and is not as precise as the popular press would have you believe.