Open catchyoo opened 1 month ago
Hello, I have a question. I am trying to extract the prediction map and GT (ground truth) map using the ADE20K dataset. However, in ADE20K, because of the zero_label, when I extract the prediction map and GT map using the code you provide, there are differences in class names, colors, etc. For example, there's an issue where the ceiling is identified as a road in the photo. I would appreciate your help on what to do.
你能提供一下 数据集的config嘛?感觉是因为 zero_label的问题