I noticed that the author discarded the outdoor window labels on the building from the original ADE 20K dataset when processing the dataset. Why is this? If I want to add this part to continue training, is there any way to quickly reorganize the original ADE 20K dataset? So that it can serve as input for internimage.
I noticed that the author discarded the outdoor window labels on the building from the original ADE 20K dataset when processing the dataset. Why is this? If I want to add this part to continue training, is there any way to quickly reorganize the original ADE 20K dataset? So that it can serve as input for internimage.