Thank you for sharing your research findings, which are impressive.
I see that the code uses a classification based loss function, while the original casmvsnet used depth difference as the loss. How much does the change in loss function affect the results?
Thank you for sharing your research findings, which are impressive.
I see that the code uses a classification based loss function, while the original casmvsnet used depth difference as the loss. How much does the change in loss function affect the results?