Skip to content

Conversation

pcuenca
Copy link

@pcuenca pcuenca commented Apr 21, 2017

Some small changes to the notebook:

  • Normalize images so their values lie in the [0, 1] range.
  • Use all the generated words if no period is part of the word list.
  • test function defaults to using a valid image path.

More details in the corresponding revision logs.

pcuenca added 3 commits April 21, 2017 13:15
…[0,1] range.

I _believe_ this is the expected range for the VGG16 model.

A better solution could be to perform a specific normalization step depending on the initial type and depth of the source image. This change is sufficient for me for now, as it allows the testing of regular JPG files whose ranges were in the [0, 255] range.
This was an usual occurrence while testing early epoch models. If the punctuation sign was not among the generated words, the `test` function only showed the first word ('a', usually).
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

Successfully merging this pull request may close these issues.

1 participant