Release 1.3.0

New features

• Support direct annotation input for COCO format (#1921)
• Action task supports multi GPU training. (#2057)
• Support storage cache in Apache Arrow using Datumaro for action tasks (#2087)
• Add a simplified greedy labels postprocessing for hierarchical classification (#2064).
• Support auto adapting batch size (#2119)
• Support auto adapting num_workers (#2165)
• Support noisy label detection for detection tasks (#2109, #2115, #2123, #2183)

Enhancements

• Make semantic segmentation OpenVINO models compatible with ModelAPI (#2029).
• Support label hierarchy through LabelTree in LabelSchema for classification task (#2149, #2152).
• Enhance exportable code file structure, video inference and default value for demo (#2051).
• Speedup OpenVINO inference in image classificaiton, semantic segmentation, object detection and instance segmentation tasks (#2105).
• Refactoring of ONNX export functionality (#2155).

Bug fixes

• Fix async mode inference for demo in exportable code (#2154)

Known issues

• OpenVINO(==2022.3) IR inference is not working well on 2-stage models (e.g. Mask-RCNN) exported from torch==1.13.1
  (working well up to torch==1.12.1) (#1906)

Release 1.2.1

Enhancements

• Upgrade mmdeploy==0.14.0 from official PyPI (#2047)
• Integrate new ignored loss in semantic segmentation (#2065, #2111)
• Optimize YOLOX data pipeline (#2075)
• Tiling Spatial Concatenation for OpenVINO IR (#2052)
• Optimize counting train & inference speed and memory consumption (#2172)

Bug fixes

• Bug fix: value of validation variable is changed after auto decrease batch size (#2053)
• Fix tiling 0 stride issue in parameter adapter (#2078)
• Fix Tiling NNCF (#2081)
• Do not skip full img tile classifier + Fix Sequencial Export Issue (#2174)

Release 1.2.0

New features

• Add generating feature cli_report.log in output for otx training (#1959)
• Support multiple python versions up to 3.10 (#1978)
• Support export of onnx models (#1976)
• Add option to save images after inference in OTX CLI demo together with demo in exportable code (#2005)
• Support storage cache in Apache Arrow using Datumaro for cls, det, seg tasks (#2009)
• Add noisy label detection for multi-class classification task (#1985, #2034)

Enhancements

• Clean up and refactor the output of the OTX CLI (#1946)
• Enhance DetCon logic and SupCon for semantic segmentation(#1958)
• Detection task refactoring (#1955)
• Classification task refactoring (#1972)
• Extend OTX explain CLI (#1941)
• Segmentation task refactoring (#1977)
• Action task refactoring (#1993)
• Optimize data preprocessing time and enhance overall performance in semantic segmentation (#2020)
• Support automatic batch size decrease when there is no enough GPU memory (#2022)

Bug fixes

• Fix backward compatibility with OpenVINO SSD-like detection models from OTE 0.5 (#1970)

Known issues

• OpenVINO(==2022.3) IR inference is not working well on 2-stage models (e.g. Mask-RCNN) exported from torch==1.13.1
  (working well up to torch==1.12.1) (#1906)

Release 1.1.2

Bug fixes

• Fix exception -> warning for anomaly dump_feature option
• Remove dataset.with_empty_annotations() to keep original input structure (#1964)
• Fix OV batch inference (saliency map generation) (#1965)
• Replace EfficentNetB0 model download logic by pytorchcv to resolve zip issue (#1967)

Release 1.1.1

Bug fixes

• Add missing OpenVINO dependency in exportable code requirement

Release 1.1.0

New features

• Add FP16 IR export support (#1683)
• Add in-memory caching in dataloader (#1694)
• Add MoViNet template for action classification (#1742)
• Add Semi-SL multilabel classification algorithm (#1805)
• Integrate multi-gpu training for semi-supervised learning and self-supervised learning (#1534)
• Add train-type parameter to otx train (#1874)
• Add embedding of inference configuration to IR for classification (#1842)
• Enable VOC dataset in OTX (#1862)
• Add mmcls.VisionTransformer backbone support (#1908)

Enhancements

• Parametrize saliency maps dumping in export (#1888)
• Bring mmdeploy to action recognition model export & Test optimization of action tasks (#1848)
• Update backbone lists (#1835)
• Add explanation for XAI & minor doc fixes (#1923)
• Refactor phase#1: MPA modules

Bug fixes

• Handle unpickable update_progress_callback (#1892)
• Dataset Adapter: Avoid duplicated annotation and permit empty image (#1873)
• Arrange scale between bbox preds and bbox targets in ATSS (#1880)
• Fix label mismatch of evaluation and validation with large dataset in semantic segmentation (#1851)
• Fix packaging errors including cython module build / import issues (#1936)

Known issues

• OpenVINO(==2022.3) IR inference is not working well on 2-stage models (e.g. Mask-RCNN) exported from torch==1.13.1
  (working well up to torch==1.12.1) (#1906)

Release v1.0.1

Enhancements

• Refine documents by proof review
• Separate installation for each tasks
• Improve POT efficiency by setting stat_requests_number parameter to 1

Bug fixes

• Fix missing classes in cls checkpoint
• Fix action task sample codes
• Fix label_scheme mismatch in classification
• Fix training error when batch size is 1
• Fix hang issue when tracing a stack in certain scenario
• Fix pickling error by Removing mmcv cfg dump in ckpt

Release v1.0.0

NOTES

OpenVINO™ Training Extensions which version 1.0.0 has been updated to include functional and security updates. Users should update to the latest version.

New features

• Adaptation of Datumaro component as a dataset interface
• Integrate hyper-parameter optimizations
• Support action recognition task
• Self-supervised learning mode for representational training
• Semi-supervised learning mode for better model quality

Enhancements

• Installation via PyPI package
• Enhance find command to find configurations of supported tasks / algorithms / models / backbones
• Introduce build command to customize task or model configurations in isolated workspace
• Auto-config feature to automatically select the right algorithm and default model for the train & build command by detecting the task type of given input dataset
• Improve documentation
• Improve training performance by introducing enhanced loss for the few-shot transfer

Bug fixes

• Fixing configuration loading issue from the meta data of the model in OpenVINO task for the backward compatibility
• Fixing some minor issues

Release v0.5.0

Added

• Add tiling in rotated detection (#1420)
• Add Cythonize AugMixAugment to accelerate data loading during training (#1478)
• Integrate ov-telemetry (#1568)

Changed

• Update OpenVINO to 2022.3 release & nncf to the pre-2.4 version (#1393)

Fixed

• Fixing indexing bugs in hierarchical label classification(#1458, #1551, #1599)
• Fixing deprecated np.bool issue from numpy==1.24.0 (#1455)
• Fixing tiling OpenVINO backward compatibility (#1516)
• Copying feature vector to resolve duplication issue (#1511)
• Some minor issues

Release v0.4.0

Added

• Model Preparation Algorithm (MPA)
  • Better saliency map support
    • Replace current saliency map generation with Recipro-CAM for cls task (#1363)
    • Class-wise saliency map generation for the detection task (#1402)
    • Align saliency map labels for all tasks (#1447)
  • Improve object counting algorithm for high-res images via image tiling
    • Add tiling module for detection, rotated detection and instance segmentation (#1200)
    • Fliter object less than 1 pixel (#1305)
    • Add tiling deployment (#1387)
  • Enabling GPU execution for exported code (#1416)
  • Add active score evaluation and probability output to the classification task (#1417)

Changed

• Normalize top-1 metrics to [0, 1] (#1394)

Fixed

• Hot-fix for Detection fix two stage error (#1433)
• Fixing ZeroDivisionError in iteration counter for detection-classification project trainings (#1449)
• Some minor issues