Permutations of key point indices of CUB and Spair-71k

[Gjergj121]

Hi. In prepare_data.py, in line 22 you state that "When an image is mirrored, any key points with left/right distinction need to be swapped. These are the permutations of key point indices that accomplishes this:". How did you calculate these permutations, and how could I calculate them for another class of the Spair-71k dataset, e.g. person?
Thank you in advance :)