You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Sep 18, 2024. It is now read-only.
Is there any reason why there isn't a quiet mode for ImageDataGenerator.flow_from_dataframe? It's not ideal to see "Found {} non-validated image filenames." for every initialization.
The text was updated successfully, but these errors were encountered:
Instead of switching the message via an ENV var I suggest adding an optional keyword argument verbose with default setting verbose=1 to these functions, such that verbose=0 suppresses these printed messages See e.g. the verbose keywords of tf.keras.Model.fit, or of the EarlyStopping or ModelCheckpoint callbacks.
According to the release notes, current Tensorflow 2.3.0 introduced a new Keras dataset generation utility (emphasis added):
image_dataset_from_directory is a utility based on tf.data.Dataset, meant to replace the legacy ImageDataGenerator. It takes you from a structured directory of images to a labeled dataset, in one function call. Note that it doesn't perform image data augmentation (which is meant to be done using preprocessing layers).
This means that I'll simply comment out the annoying print command locally for now.
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Is there any reason why there isn't a quiet mode for ImageDataGenerator.flow_from_dataframe? It's not ideal to see "Found {} non-validated image filenames." for every initialization.
The text was updated successfully, but these errors were encountered: