elnino9ykl / pass Goto Github PK

Hi @elnino9ykl,
thank you for sharing the code. I find it very valuable.

I try to run the script main.py under train to train the model.
However, I don't know how to prepare the dataset folder for datadir argument.
When I read the code and the paper, I guess that you use a subset of 27 classes out of 66 classes in Vista dataset.
However, the code only supports Cityscapes dataset as written in cityscapes class in piwise/dataset.py.

Could you please show me the proper way to prepare the dataset folder for datadir argument?
If possible, could you please share the script to convert Mapillary Vista to your desire format (folder structure, select 27 classes only, ...)?

Thank you.

Can you please provide google drive link for the dataset with annotations for training and validation as baidu is pretty slow?