@knud @franco_vazza
Thank you for the discussion! Actually, the last classification presented in our paper is done without human interference, so probably it can be used for incoming large piles of data!