Closed sm-potter closed 10 months ago
This isn't an issue but I am just wondering how the model was trained when clouds or other no data pixels were present? Is there a specific value we should assign to them to align with the pre-training?
Closing as I found here for burn scars at least, no data is -1.https://huggingface.co/datasets/ibm-nasa-geospatial/hls_burn_scars
This isn't an issue but I am just wondering how the model was trained when clouds or other no data pixels were present? Is there a specific value we should assign to them to align with the pre-training?