Add multi-label classification dataset and metric #1572
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Add MultiLabel annotation support - Refactor de/serialize annotation with AnnotationRaw - Add ImageFolderDataset::with_items methods
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1572 +/- ##
==========================================
- Coverage 86.53% 86.34% -0.19%
==========================================
Files 684 687 +3
Lines 78248 78685 +437
==========================================
+ Hits 67713 67943 +230
- Misses 10535 10742 +207 ☔ View full report in Codecov by Sentry. |
Comment on lines
111
to
123
| fn bin_config() -> bincode::config::Configuration { | ||
| bincode::config::standard() | ||
| } | ||
|
|
||
| fn encode(&self) -> Vec<u8> { | ||
| bincode::serde::encode_to_vec(self, Self::bin_config()).unwrap() | ||
| } | ||
|
|
||
| fn decode(annotation: &[u8]) -> Self { | ||
| let (annotation, _): (AnnotationRaw, usize) = | ||
| bincode::serde::decode_from_slice(annotation, Self::bin_config()).unwrap(); | ||
| annotation | ||
| } |
There was a problem hiding this comment.
We use the serialization for what exactly?
There was a problem hiding this comment.
We decided on having annotations as bytes
struct ImageDatasetItemRaw {
/// Image path.
image_path: PathBuf,
/// Image annotation.
/// The annotation bytes can represent a string (category name) or path to annotation file.
annotation: Vec<u8>,
}But now that you mention it... I don't see any need for serialization just to have bytes 😅 we could simply change the annotation type in ImageDatasetItemRaw to the AnnotationRaw enum. And scrap the encode/decode.
Probably needed another coffee when I went over this part ☕
antimora
requested changes
Apr 4, 2024
antimora
approved these changes
Apr 4, 2024
nathanielsimard
approved these changes
Apr 5, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Pull Request Template
Checklist
run-checks allscript has been executed.Related Issues/PRs
Progress towards #1526.
Fine-tuning example is almost complete (locally).
Changes
Added
ImageFolderDataset::new_multilabel_classification_with_itemsandHammingScoremulti-label accuracy metricAnnotation::MultiLabel(Vec<usize>)for multi-label classificationAnnotationRawenum to de/serialize different supported annotation types withbincodeImageFolderDatasetnew methods to usewith_itemsHammingScoremetric andMultiLabelClassificationOutputto handle multi-label outputsTesting
New unit tests for dataset methods and hamming score metric.