May I ask how you prepare the OVS_datasets? The patch size in your provided datasets is very small, which is 256x256, and not all patches cropped from the original image are included in you datasets. Could you share the procedure and principles you follow to get the OVS_datasets?
Hi, I followed the paper few-shot segmentation benchmark paper [Scale-aware detailed matching for few-shot aerial image semantic segmentation] to prepare the iSAID dataset.
Hi,
May I ask how you prepare the OVS_datasets? The patch size in your provided datasets is very small, which is 256x256, and not all patches cropped from the original image are included in you datasets. Could you share the procedure and principles you follow to get the OVS_datasets?
Regards, Shiqi