finetune on fingerprint dataset
whqwill opened this issue · 2 comments
whqwill commented
Hi,
I try to use the framework to finetune on fingerprint dataset, but the loss is very high and after several hours it is still very high, the loss dose not change. But I tried some other datasets, it works week. I want to know if this model is not suitable for the fingerprint.
tynguyen commented
Hi there,
There are a couple of possible reasons:
1. The displacement of a pixel between the two images is too big? Make sure
that it should not > 30%
2. The transformation is not homography?
3. There are distortions in parts of the images?
It's hard to tell correctly what is the reason without seeing the actual
images though.
…On Mon, Nov 4, 2019 at 9:21 PM whqwill ***@***.***> wrote:
Hi,
I try to use the framework to finetune on fingerprint dataset, but the
loss is very high and after several hours it is still very high, the loss
dose not change. But I tried some other datasets, it works week. I want to
know if this model is not suitable for the fingerprint.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#14?email_source=notifications&email_token=ABKAZDMJGI54B3BGA3ZDF3LQSDKA3A5CNFSM4JI4U3KKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HWYSVPA>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABKAZDPAFWH4VMODR4IJL6DQSDKA3ANCNFSM4JI4U3KA>
.
whqwill commented
I made the pictures by gen_synthetic_data.py, the pictures are like this:
but I tried other pictures like this, it works very well:
