about loss function

[cuongdxk57]

i am confused about loss function in the paper, when estimated character bounding boxes is low mean the accuracy of model is bad, confidence score is low and the loss is low too.

[uname0x96]

When you predict too less char the value computed as: will high. And the confidence score they want it approximately 1 because when it is 1 that mean weakly suplearning is good for task split word to char.