What is the additional cityscapes `train_extra` data needed for SOTA training?

[the-joy-of-data]

Hello,

Thank you for sharing your work. Very intuitive README and code structure is very well documented.

I ran some eval tests as mentioned in the README and they went well, I get the exact same IoU of 86.92 for eval_cityscapes.yml

Next, I tried train_cityscapes.yml. I needed to bump down the crop size to 512, 1024 because I was running out of memory, and that worked as well. Please comment if this reduction in crop_size could have an impact on IoU accuracy.

My main question, however, is when I try to train train_cityscapes_sota.yml it breaks while looking for this data folder. leftImg8bit_trainextra/leftImg8bit/train_extra

Could you please share what is this additional / extra data needed from cityscapes to train the sota run? Or was this just a different split that you tried from the original dataset?

[the-joy-of-data]

found it in https://github.com/NVIDIA/semantic-segmentation/blob/main/PREPARE_DATASETS.md, thanks