Ctrl+K
  • GitHub
  • GitHub

Release Notes#

v1.10.0 (2024 Q4)

New features#

  • Support KITTI 3D format

  • Add PseudoLabeling transform for unlabeled dataset

Enhancements#

  • Raise an appropriate error when exporting a datumaro dataset if its subset name contains path separators.

  • Update docs for transform plugins

  • Update ov ir model for explorer openvino launcher with CLIP ViT-L/14@336px model

  • Optimize path assignment to handle point cloud in JSON without images

  • Set TabularTransform to process clean transform in parallel

Bug fixes#

  • Fix datumaro format to load visibility information from Points annotations

v1.9.1 (2024 Q3)#

Enhancements#

  • Support multiple labels for kaggle format

  • Use DataFrame.map instead of DataFrame.applymap

Bug fixes#

  • Fix StreamDataset merging when importing in eager mode

v1.9.0 (2024 Q3)#

New features#

  • Add a new CLI command: datum format

  • Support language dataset for DmTorchDataset

Enhancements#

  • Change _Shape to Shape and add comments for subclasses of Shape

Bug fixes#

  • Fix KITTI-3D importer and exporter

v1.8.0 (2024 Q3)#

New features#

  • Add TabularValidator

  • Add Clean Transform for tabular data type

Enhancements#

  • Set label name with parents to avoid duplicates for AstypeAnnotations

  • Pass Keyword Argument to TabularDataBase

Bug fixes#

  • Preserve end_frame information of a video when it is zero.

  • Changed the Datumaro format to ensure exported videos have relative paths and to prevent the same video from being overwritten.

v1.7.0 (2024 Q2)#

New features#

  • Add ann_types property for dataset

  • Add AnnotationType.rotated_bbox for oriented object detection

  • Add DOTA data format for oriented object detection task

  • Add AstypeAnnotations Transform

Enhancements#

  • Fix ambiguous COCO format detector

  • Get target information for tabular dataset

  • Add ExtractedMask and update importers who can use it to use it

v1.6.1 (2024.05)#

Enhancements#

  • Prevent AcLauncher for OpenVINO 2024.0

Bug fixes#

  • Modify lxml dependency constraint

  • Fix CLI error occurring when installed with default option only

  • Relax Pillow dependency constraint

  • Modify Numpy dependency constraint

  • Relax old pandas version constraint

v1.6.0 (2024.04)#

New features#

  • Changed supported Python version range (>=3.9, <=3.11)

  • Support MMDetection COCO format

  • Develop JsonSectionPageMapper in Rust API

  • Add Filtering via User-Provided Python Functions

  • Remove supporting MacOS platform

  • Support Kaggle image data (KaggleImageCsvBase, KaggleImageTxtBase, KaggleImageMaskBase, KaggleVocBase, KaggleYoloBase)

  • Add __getitem__() for random accessing with O(1) time complexity

  • Add Data-aware Anchor Generator

  • Support bounding box import within Kaggle extractors and add KaggleCocoBase

Enhancements#

  • Optimize Python import to make CLI entrypoint faster

  • Add ImageColorScale context manager

  • Enhance visualizer to toggle plot title visibility

  • Enhance Datumaro data format detect() to be memory-bounded and performant

  • Change RoIImage and MosaicImage to have np.uint8 dtype as default

  • Enable image backend and color channel format to be selectable

  • Boost up CityscapesBase and KaggleImageMaskBase by dropping np.unique

  • Enhance RISE algortihm for explainable AI

  • Enhance explore unit test to use real dataset from ImageNet

  • Fix each method of the comparator to be used separately

Bug fixes#

  • Fix wrong example of Datumaro dataset creation in document

  • Fix wrong command to install datumaro from github

  • Update document to correct wrong datum project import command and add filtering example to filter out items containing annotations.

  • Fix label compare of distance method

  • Fix Datumaro visualizer’s import errors after introducing lazy import

  • Fix broken link to supported formats in readme

  • Fix Kinetics data format to have media data

  • Handling undefined labels at the annotation statistics

  • Add unit test for item rename

  • Fix a bug in the previous behavior when importing nested datasets in the project

  • Fix Kaggle importer when adding duplicated labels

  • Fix input tensor shape in model interpreter for OpenVINO 2023.3

  • Add default value for target in prune cli

  • Remove deprecated MediaManager

  • Fix explore command without project

v1.5.2 (2024.01)#

Enhancements#

  • Add memory bounded datumaro data format detect

  • Remove Protobuf version limitation (<4)

v1.5.1 (2023.11)#

Enhancements#

  • Enhance Datumaro data format stream importer performance

  • Change image default dtype from float32 to uint8

  • Add comparison level-up doc

  • Add ImportError to catch GitPython import error

Bug fixes#

  • Modify the draw function in the visualizer not to raise an error for unsupported annotation types.

  • Correct explore path in the related document.

  • Fix errata in the voc document. Color values in the labelmap.txt should be separated by commas, not colons.

  • Fix hyperlink errors in the document.

  • Fix memory unbounded Arrow data format export/import.

  • Update CVAT format doc to bypass warning.

v1.5.0 (2023.09)#

New features#

  • Add tabular data import/export

  • Support video annotation import/export

  • Add multiframework (PyTorch, Tensorflow) converter

  • Add SAM OVMS and Triton server Docker image builders

  • Add SAMBboxToInstanceMask transform

  • Add ConfigurableValidator

Enhancements#

  • Enhance ClassificationValidator for multi-label classification datasets with label_groups

  • Replace Roboflow xml.etree with defusedxml

  • Define GroupType with IntEnum for, where 0 is EXCLUSIVE

  • Add Rust API to optimize COCOPageMapper performance

  • Support a dictionary input in addition to a single image input for the model launcher to support Segment Anything Model

  • Remove deprecates announced to be removed in 1.5.0

  • Add multi-threading option to ModelTransform and SAMBboxToInstanceMask

Bug fixes#

  • Fix bugs for Tile transform

  • Disable Roboflow Tfrecord format when Tensorflow is not installed

  • Raise VcsAlreadyExists error if vcs directory exists

v1.4.1 (2023.07)#

Bug fixes#

  • Report errors for COCO (stream) and Datumaro importers

v1.4.0 (2023.07)#

New features#

  • Add documentation and notebook example for Prune API

  • Changed supported Python version range (>=3.8, <=3.11)

  • Migrate OpenVINO v2023.0.0

  • Add Roboflow data format support (COCO JSON, Pascal VOC XML, YOLOv5-PyTorch, YOLOv7-PyTorch, YOLOv8, YOLOv5 Oriented Bounding Boxes, Multiclass CSV, TFRecord, CreateML JSON)

  • Add MissingAnnotationDetection transform

  • Add OVMSLauncher

  • Add Prune API

  • Add TritonLauncher

  • Migrate DVC v3.0.0

  • Stream dataset import/export

  • Support mask annotations for CVAT data format

Enhancements#

  • Support list query for explorer

  • update contributing.md

  • Update 3rd-party.txt for release 1.4.0

  • Give notice that the deprecation works will be done in datumaro==1.5.0

  • Unify COCO, Datumaro, VOC, YOLO importer/exporter progress reporter descriptions

  • Enhance import performance for built-in plugins

  • Change default dtype of load_image() to np.uint8

  • Add OTX ATSS detector model interpreter & refactor interfaces

  • Refactor Launcher and ModelInterpreter

  • Add CVAT data format document

  • Reduce peak memory usage when importing COCO and Datumaro formats

  • Enhance the error message for datum stats to be more user friendly

  • Refactor dataset.py to seperate DatasetStorage

Bug fixes#

  • Create cache dir under only writable filesystem

  • Fix: Dataset infos() can be broken if a transform not redefining infos() is stacked on the top

  • Fix warnings in test_visualizer.py

  • Fix LabelMe data format

  • Prevent installing protobuf>=4

  • Fix UnionMerge

v1.3.2 (2023.06)#

Enhancements#

  • Let CocoBase continue even if an InvalidAnnotationError is raised

Bug fixes#

  • Install dvc version to 2.x

  • Replace np.append() in Validator

v1.3.1 (2023.05)#

Bug fixes#

  • Fix Cityscapes format mis-detection problem

v1.3.0 (2023.05)#

New features#

  • Add CocoRoboflowImporter

  • Add SynthiaSfImporter and SynthiaAlImporter

  • Add intermediate skill document for filter

  • Add VocInstanceSegmentationImporter and VocInstanceSegmentationExporter

  • Add Segment Anything data format support

  • Add Correct transformation

  • Add ReindexAnnotations transform

Enhancements#

  • Use autosummary for fully-automatic Python module docs generation

  • Enrich stack trace for better user experience when importing

  • Save and load hashkey for explorer

  • Add MOT and MOTS data format documents

  • Improve RemoveAnnotations to remove specific annotations with ids

  • Add Jupyter notebook example of noisy label detection for detection tasks

  • Add Juypter notebook examples for importing/exporting detection and segmentation data

Bug fixes#

  • Fix Mapillary Vistas data format

  • Fix bytes property returning None if function is given to data

  • Fix Synthia-Rand data format

  • Fix person_layout categories and action_classification attributes in imported Pascal-VOC dataset

  • Drop a malformed transform from StackedTransform automatically

  • Fix Cityscapes to drop ImgsFine directory

v1.2.1 (2023.05)#

Bug fixes#

  • Fix project level CVAT for images format import

  • Fix an info message when using the convert CLI command with no args.input_format

  • Fix media contents not returning bytes in arrow format

v1.2.0 (2023.04)#

New features#

  • Add Skill Up section to documentation

  • Add LossDynamicsAnalyzer for noisy label detection

  • Add Apache Arrow format support

  • Add sort transform

Enhancements#

  • Add multiprocessing to DatumaroBinaryBase

  • Refactor merge code

  • Refactor download CLI commands

  • Refactor CLI commands w/ and w/o project

  • Refactor Media to be initialized from explicit sources

  • Refactor hl_ops.py

  • Add tfds:uc_merced and tfds:eurosat download

  • Migrate documentation framework to Sphinx

  • Update merge tutorial for real life usecase

  • Abbreviate “detect-format” to “detect” for prettifying

Bug fixes#

  • Add UserWarning if an invalid media_type comes to image statistics computation

  • Fix negated is_encrypted

  • Save extra images of PointCloud when exporting to datumaro format

  • Fix log issue when importing celeba and align celeba dataset

v1.1.0 (2023.03)#

New features#

  • Add with_subset_dirs decorator (Add ImagenetWithSubsetDirsImporter)

  • Add CommonSemanticSegmentationWithSubsetDirsImporter

  • Add DatumaroBinary format

  • Add Searcher CLI documentation

  • Add version to dataset exported as datumaro format

  • Add Ava action data format support

  • Add Shift Analyzer (both covariate and label shifts)

  • Add YOLO Loose format

  • Add Ultralytics YOLO format

Enhancements#

  • Refactor Datumaro format code and test code

Bug fixes#

  • Fix image filenames and anomaly mask appearance in MVTec exporter

  • Fix CIFAR10 and 100 detect function

  • Fix celeba and align_celeba detect function

  • Choose the top priority detect format for all directory depths

  • Fix MVTec format detect function

  • Fix wrong __len__() of Subset when the item is removed

  • Fix mask visualization bug

v1.0.0 (2023.02)#

New features#

  • Add Data Explorer

  • Add Ellipse annotation type

  • Add MVTec anomaly data support

Enhancements#

  • Refactor existing tests

  • Raise ImportError on importing malformed COCO directory

  • Remove the duplicated and cyclical category context in documentation

Bug fixes#

  • Fix for importing CVAT image 1.1 data format exported to project level

  • Fix a problem on setting log-level via CLI

  • Fix code format with the latest black==23.1.0

  • Fix ‘Explain command cannot find the model’

  • Fix a problem found on model remove CLI command

Note

About the release of the developed version can be read in the CHANGELOG.md of the develop branch.

On this page
Show Source