You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Could you please tell me the training procedure for Indria dataset from scratch? I have aerial images and corresponding masks in .png format right now.
The text was updated successfully, but these errors were encountered:
So for the Inria dataset there are 2 possibilities:
Download OSM annotations for the images. In my case I also fixed the alignment of those annotations with a previous method of mine. To use this version of the dataset, reference the dataset_params.inria_dataset_osm_aligned.json config file.
Convert the available ground truth masks with the frame field polygonization method. To use this version of the dataset, reference the dataset_params.inria_dataset_polygonized.json config file.
I added details for this in the main README as well as a link to download all those different ground truth files so you can get started right away!
Could you please tell me the training procedure for Indria dataset from scratch? I have aerial images and corresponding masks in .png format right now.
The text was updated successfully, but these errors were encountered: