Filtering by subfolder option in parse_folder script #215
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This pull request is the result of the discussion in #212.
It is aimed at handling the following import case when parsing data folders.
Simple example: there are images from two categories and for every image we create augmented versions by mirroring resulting in a folder structure comparable to:
Currently, the structure is just flattened and then splitted into train and validation datasets. So it could happen that this split is chosen:
Training:
Validation:
That is not the desired result because it mixes data that originated from the same source image in training and validation. What you would want is:
Training:
Validation:
A new argument called
--split_by_subfolder
has been added tohttps:/crohkohl/DIGITS/blob/split_by_subfolder/tools/parse_folder.py#L524
which leads to that behaviour. The data is first grouped by the deepest sub-folder name in a dictionary - then divided into train / val / test and finally all group items are added to the image lists.