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
Thanks for sharing the source codes.
When opt.arb_sample is set to be 1 (flexible sampling), the opt.crop_size is 0. However, when opt.arb_sample is set to be 0 (fixed sampling), the opt.crop_size is opt.psv_range * (opt.angular_out - 1). I did not find the reason for these two different settings in your paper. Can you please help me with it?
The text was updated successfully, but these errors were encountered:
As PSV has zero boundaries which will make the training unstable, we used non-zero crop_size for the fixed sampling. This influence is not obvious when arb_sample, so we did not apply the cropping.
Thanks for sharing the source codes.
When opt.arb_sample is set to be 1 (flexible sampling), the opt.crop_size is 0. However, when opt.arb_sample is set to be 0 (fixed sampling), the opt.crop_size is opt.psv_range * (opt.angular_out - 1). I did not find the reason for these two different settings in your paper. Can you please help me with it?
The text was updated successfully, but these errors were encountered: