HighRes-net
HighRes-net copied to clipboard
Calculation of val_score
So val_score is calculated using shift_cPSNR.
Some things need explanation:
- Why best_score starts from 100? and why shift_cPSNR is reduced from that value?
- Why val_score is normalized by size of validation dataset after looping over srs.shape values? What is its effect? Being reduced from 100 to begin with?
- Why the final score (after calculation) ends up being negative? What is this metric?
Thanks