Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Unwanted symbols appear on generated images #72

Open
DLumi opened this issue Oct 8, 2024 · 0 comments
Open

Unwanted symbols appear on generated images #72

DLumi opened this issue Oct 8, 2024 · 0 comments

Comments

@DLumi
Copy link

DLumi commented Oct 8, 2024

I have my corpus set up like so. By design it contains only single characters (but I specified that in the config anyway).

corpus:
  weights: [1]
  args:
    # length augmentable corpus
    - paths: [resources/corpus/mrz.txt]
      weights: [1]
      min_length: 1
      max_length: 1
      textcase: [upper]
      augmentation: 0
      augmentation_length: [0]

As for char augmentable corpus, I tried fiddling with the settings as well, but it does not seem to help.

So what happens is sometimes I get longer strings like so:

images/0/0.jpg	Z
images/0/1.jpg	O
images/0/4.jpg	4
images/0/3.jpg	4
images/0/5.jpg	N
images/0/8.jpg	A
images/0/6.jpg	P
images/0/10.jpg	9
images/0/2.jpg	HJ4I
images/0/7.jpg	K
images/0/11.jpg	I
images/0/9.jpg	E
images/0/13.jpg	S
images/0/15.jpg	M
images/0/14.jpg	V
images/0/16.jpg	Q
images/0/19.jpg	N
images/0/18.jpg	X
images/0/21.jpg	W
images/0/20.jpg	2
images/0/17.jpg	A
images/0/23.jpg	Q
images/0/22.jpg	M
images/0/25.jpg	K
images/0/27.jpg	U
images/0/24.jpg	X
images/0/26.jpg	YXN2MQ<7NP51CNUYTFI

image

And this is despite the fact that this simply should not happen. I'm expecting to get single characters only.

Those long strings are few in numbers, and I could filter them out. But I'd rather not generate them at all.
So is it me not setting up something properly, or is it a synthtiger bug?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant