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

New Training Data #3383

Closed
DimChik2903 opened this issue Nov 13, 2024 · 4 comments
Closed

New Training Data #3383

DimChik2903 opened this issue Nov 13, 2024 · 4 comments

Comments

@DimChik2903
Copy link

Files

Hello! Thanks for your work! I have problems recognizing the numbers 0,6,8,9. I am attaching the archive.

my_digits.zip

@caco3
Copy link
Collaborator

caco3 commented Dec 1, 2024

They look like digits we already have in the trainings data: https://jomjol.github.io/neural-network-digital-counter-readout/

Are you sure they are needed?

@DimChik2903
Copy link
Author

DimChik2903 commented Dec 19, 2024

I constantly have mistakes with round numbers: 0,6,8,9. They always get confused. If there is a way to fix this error without training, I would appreciate a hint.

image

@SybexX
Copy link
Collaborator

SybexX commented Dec 19, 2024

This will be due to the reflection, try to eliminate it. So you shouldn't have any problems with the first, second and fifth digits, but you should have problems with the rest.
Unbenannt

@jomjol
Copy link
Owner

jomjol commented Jan 5, 2025

It doesn't make sense to compensate reflections by training them into the neural network. This will make other detections worse.
So please remove the reflections, this kind of numbers is in the training set already.
Thx, jomjol

@jomjol jomjol closed this as completed Jan 5, 2025
@caco3 caco3 closed this as not planned Won't fix, can't repro, duplicate, stale Jan 5, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants