Training the model on another dataset

Question

Training the model on another dataset

Closed this issue 2 months ago · 2 comments

I want to train the model on another dataset similar to cityscapes but with other labels.
How can I do this? Are there any challenges?

Answer 1 · 2024-05-28T21:16:06.000Z

Hi @yassine9666 ,
Thank you for the interest! As mentioned in the implementation details of the method section, you would need to initialize the discriminator with pretrained segmenter weights. The pretrained UperNet weights are uploaded here, which are originally from semantic-segmentation-pytorch with semantic classes re-matched based on the training dataset

Answer 2 · 2024-06-16T09:15:22.000Z

Hi @yassine9666 , I have also updated README here. Hopefully, it can provide some guidelines regarding training on a new dataset.