Confusions about the Preparation for 2018 dataset

[lofrienger]

Hi, thanks for your great work！I am preprocessing 2018 dataset according to https://github.com/BCV-Uniandes/ISINet/blob/main/data/README.md
However, some details are not very clear. I met some issues when I tried to get the coco format annotation for 2018 dataset.

I have already downloaded original 2018 dataset (part-wise annotation) and your dataset (intrument-wise annotation). And I put both dataset in the same folder. Although I noticed you mentioned "use the 2018 labels compatible with the 2017 dataset (final_labels_2017.json).", I don't know how to prepare it. When I run organize2018.py, it requires 'instrument_labels_2017' folder.
May I know how to solve it?
Can you kindly show the dataset folder structure in GitHub, it would be very helpful!

https://github.com/BCV-Uniandes/ISINet/blob/main/data/organize2018.py line 164
instr_label_folder = osp.join(args.data_dir, data_f, in_f, 'instrument_labels_2017')

Thanks for your help!