Hi - the public leaderboard is the preferred way to evaluate on that test set: It provides reviewers of journals & conferences an independent trusted evaluation, encourages respecting real test data, reduces risk of blurring lines between test- and training data leading to biased results, and discourages ‘wacky boosting’.
If there are any requirements of you not covered by the platforms’ functionality, and need anything beyond please contact the ISIC team via challenge@isic-archive.com