The answer depends on what you mean by "count". If we consider whether they should be counted as citations: yes definitely. Afer all, the citations reflect how your science is read and then cited by others. It is then clear that these "count".
A problem occurs when using the citation counts provided by different services such as Researcher ID through Web of Science. Then it is not certain that the "in press" version is counted. I, for example have papers where in press citations exist. In Web of Science, I can actually see them but the connection is not automatically done so I would have to add them to the appropriate paper count manually. I know it is possible to write to Web Of Science and have references corrected or coupled. I, for example have another paper that occurs in four posts because people have written the wrong volume number, the wrong year of publication etc. Since I am senior and these extra citations are not critical to me I have not tried to get them corrected but they can of course be more important in an early career case.
Other services such as Google Scholar may work differently. I have opted for providing measures from services to which I can link and thereby have others verify my numbers. So in the end this becomes a technical problem, the citations certainly count.