I have used s3dis dataset for trained model, but generated strange pergroup_thres.txt and mingroupsize.txt,which is shown below :
use them in generate_results.py, i get a very pool result in semantic segment and instance segment. so, i don't konw why, can you help me solve this problem?