I was noticing that OP2MapImager was being flagged as a C project by GitHub, which is incorrect since most all the source code is C++.
Turns out GitHub was counting all the FreeImage code files in the language count, which skewed the results heavily.
While not a huge deal, I thought this was worth fixing so when someone stumbles on a project they don't dismiss it since it appears to be written mostly in C or some other language, when it actually is not.
I added the following info to the .gitattributes file, which fixed the repo to list as only a C++ project:
$ cat .gitattributes
FreeImage/* linguist-vendored
ReadMe.txt linguist-documentation
So, lesson here is if you are going to include library code/header files, etc in your repository, you can flag them as linguist-vendored so they don't skew your language stats.
Also, if you have a significant number of documentation files in .txt, .xml, .md, etc, you can flag them as linguish-documentation and also exclude them from the count.
Although this all is probably not worth the time to curate unless something major is skewing the results as a single ReadMe.txt isn't going to throw off the language stats. (This might be a bad example as I don't know if GitHub auto-filters out ReadMe files from stats).
https://github.com/github/linguist#using-gitattributes-Brett