It uses the NYU pre-training model to verify the semantic segmentation effect on the Replica dataset

[jmwang0117]

Appreciating your excellent work !!!
In order to perform semantic segmentation on my Replica dataset using your pre-trained model from NYU, I have run into a few issues. I would appreciate your assistance and response.
The Replica dataset contains 900 images and 88 class labels. I changed the files under the nyuv2 data with the RGB, labels_88, and labels_88_colored files from Replica. The program displays an error when I run it.
How can I change the program so that it can adapt to the new dataset, even if some classes won't be split, since the class specified in your program is 40 and mine is 88? ?

[danielS91]

Looks like that you did not change the number of classes or did not adapted the colormaps.
I would recommend to create a new dataset class instead of changing the existing NYUv2 class - see our abstract base class for further details and function/properties that need to implemented in a derived class.
However, if your dataset provides annotations for multiple tasks, I would recommend our follow-up work as starting point.