loss = 0.0000
benlee73 opened this issue · comments
I don't know why loss is zero.
The training worked well with the icdar2015 dataset, but when I used my other dataset image and ground truth, I got this result.
The format of ground truth is the same as icdar2015.
How can I fix the problem?
For reference, my image size is about 601x792.
And i'm not change another config or something.