The fundamental question here is: Can we "read" emotions in a perceiver-independent, objective way?
The answer, I'd venture, is: We don't really know (although, currently, we cannot do this with great reliability or specificity).
It would be painstaking to go through the history of emotion research and to go over every emotion theory that supposedly has a definitive answer to this question. Fortunately, a great overview of this history can be found here.
It would also be a pain to go through every emotion perception study, in cultures near and far, every psychophysiology study, every behavioral or neuroimaging study, and the many (sometimes quite giant) limitations of this literature. Fortunately, you can find meta-analyses on these topics, which mostly suggest we cannot "recognize" contempt, or any other emotion, with notable reliability or specificity (although we can "perceive" it; e.g., 1, 2, 3, 4.)
So, it's hard to answer this question. We might be able to one day, but we are very, very, very, very far away from sampling the dynamic and highly variable range of potential expressions of "contempt," let alone any other emotion (no study has actually tried to do this ever). Moreover, quite a few emotion theories do not recognize contempt as universal (see here for a recent overview).
If the above doesn't confuse you enough, I'll pull in the picture I pointed to in the comments. Same "disgust" face, different emotion. Well, what about contempt?
I realize my answer is pretty unsatisfying, but if you read a great deal of the emotion literature, you might just arrive to the same conclusion.