anmol
anmol
Try changing the loss function from to **GDSC**
Hi I was also facing the same issue. But when I changed the loss function from Dice to **GDSC** I was able to train my network and loss also started...
Hi, Total text gives character level mask but it does not gives coordinates of every character in the image. But you have mentioned character level annotation as yes.
Hi, I was also facing the same problem. I changed the conf_thres and increased to 0.6 from 0.1. With the confidence value of 0.1 model was predicting around 51000 bounding...
I'm also facing the same issue..