Closed hukelin closed 1 year ago
Hello!
The model is trained from scratch (e.g. random initialisation), so you don't need a pre-trained model.
Hope this helps,
Devis
Thank you very much. Your work is fantastic and very useful.
I have another question, there is a buffer parameter, I don't know what this parameter means, does it indicate the width of the generated line? What is the appropriate setting for this value?
Can you specify which notebook/workflow you refer to? (I think there are multiple buffer parameter along the entire pipeline)
Hello!
The model is trained from scratch (e.g. random initialisation), so you don't need a pre-trained model.
Hope this helps,
Devis
Skimping through the notebook, it seems that we'll have to provide our own "GSAA" data to train the model, correct? If so, is there any chance you can provide us with the pre-trained model ( trained on Lithuania IIRC) so we can jump to predictions directly?
Thanks in advance.
cc: @aghand0ur
If so, is there any chance you can provide us with the pre-trained model ( trained on Lithuania IIRC) so we can jump to predictions directly?
Hi @Doomski99,
we cannot share the model, but feel free to reach out to "eoresearch at sinergise dot com" for commercial collaboration on your area of interest.
Closing issue due to inactivity.
Hello, how can I get the weight file of the pre-training model?