After some weeks of fiddling with opencv chessboard functions, I finally managed to detect some information on a modified board. I do still need to address some small mistakes that are related to color detection. These should be ironed out in the next couple of weeks. For further detail on what the project is targeting and how the project is developed please go to the github repository and read the project description.
My next move regarding detection accuracy is to create a dynamic color range histogram. This will entail having 6 known squares with the colors that I want to detect and from these, calculate some simple statistical values (mean, median, quartile). Finally, with the calculated statistical information, I can create color ranges that will then define the detection algorithm. I hope that this will be enough to get rid of some errors I have seen in the resulting bits.