Dear author,
Excuse you again. In your paper, you mentioned that you reproduced the DAFormer-based domain adaptation experiment from Potsdam to Vaihingen. I am also reproducing this experiment. The DAFormer-based domain adaptation experiment is also reproduced, but I will reproduce the process again. Some problems were encountered, especially in obtaining 'sample_class_stats.json'. I used the convert_datasets/gta.py file given by the DAFormer code, but I found that id_to_trainid was not given in the data preprocessing files of Potsdam and Vaihingen. I tried for a long time without success. Could the author give some suggestions? Thanks again!
Dear author, Excuse you again. In your paper, you mentioned that you reproduced the DAFormer-based domain adaptation experiment from Potsdam to Vaihingen. I am also reproducing this experiment. The DAFormer-based domain adaptation experiment is also reproduced, but I will reproduce the process again. Some problems were encountered, especially in obtaining 'sample_class_stats.json'. I used the convert_datasets/gta.py file given by the DAFormer code, but I found that id_to_trainid was not given in the data preprocessing files of Potsdam and Vaihingen. I tried for a long time without success. Could the author give some suggestions? Thanks again!