About Results Replication #14
Description
After I got the results using the model you provided, I used the evaluation method locally with the GDD dataset and got values different from the paper,
(iou : 0.8821552176210746 PA : 0.9443971894745149 Fb : 0.9288173651509016 MAE : 0.05560281056225196 BER : 5.37304330010211)
I have reproduced this evaluation method on GDNet, and I can reproduce the evaluation results of GDNet, but I cannot reproduce the results of EBLnet, is there any problem?