Release v1.4.1

Enhancements

• Update the README file in exportable code (#2411)

Bug fixes

• Fix broken links in documentation (#2405)

Full Changelog: 1.4.0...1.4.1

Release v1.4.0

New features

• Support encrypted dataset training (#2209)
• Add custom max iou assigner to prevent CPU OOM when large annotations are used (#2228)
• Auto train type detection for Semi-SL, Self-SL and Incremental: "--train-type" now is optional (#2195)
• Add per-class XAI saliency maps for Mask R-CNN model (#2227)
• Add new object detector Deformable DETR (#2249)
• Add new object detector DINO (#2266)
• Add new visual prompting task (#2203, #2274, #2311, #2354, #2318)
• Add new object detector ResNeXt101-ATSS (#2309)

Enhancements

• Introduce channel_last parameter to improve the performance (#2205)
• Decrease time for making a workspace (#2223)
• Set persistent_workers and pin_memory as True in detection task (#2224)
• New algorithm for Semi-SL semantic segmentation based on metric learning via class prototypes (#2156)
• Self-SL for classification now can recieve just folder with any images to start contrastive pretraining (#2219)
• Update OpenVINO version to 2023.0, and NNCF verion to 2.5 (#2090)
• Improve XAI saliency map generation for tiling detection and tiling instance segmentation (#2240)
• Remove CenterCrop from Classification test pipeline and editing missing docs link(#2375)
• Switch to PTQ for sseg (#2374)

Bug fixes

• Fix the bug that auto adapt batch size is unavailable with IterBasedRunner (#2182)
• Fix the bug that learning rate isn't scaled when multi-GPU trianing is enabled(#2254)
• Fix the bug that label order is misaligned when model is deployed from Geti (#2369)
• Fix NNCF training on CPU (#2373)
• Fix H-label classification (#2377)
• Fix invalid import structures in otx.api (#2383)
• Add for async inference calculating saliency maps from predictions (Mask RCNN IR) (#2395)

Known issues

• OpenVINO(==2023.0) IR inference is not working well on 2-stage models (e.g. Mask-RCNN) exported from torch==1.13.1

New Contributors

• @Wovchena made their first contribution in #1952
• @QSmally made their first contribution in #2046

Full Changelog: v1.0.0...1.4.0

Release v1.4.0rc4

What's Changed

• Fix invalid import structures in otx.api by @harimkang in #2383
• Update for 1.4.0rc4 by @yunchu in #2385

Full Changelog: 1.4.0rc3...1.4.0rc4

Release v1.4.0rc3

What's Changed

• Fix NNCF training on CPU by @sovrasov in #2373
• Align label order between Geti and OTX by @eunwoosh in #2369
• Remove CenterCrop from Classification test pipeline and editing missing docs link by @sungmanc in #2375
• Fix H-label classification by @sungmanc in #2377
• Update for release 1.4 by @yunchu in #2380
• Switch to PTQ for sseg by @sovrasov in #2374

Full Changelog: 1.4.0rc2...1.4.0rc3

Release v1.4.0rc2

What's Changed

• Skip openvino integration test by @sungchul2 in #2323
• Fix import dm.DatasetSubset by @vinnamkim in #2324
• Fix semantic segmentation soft prediction dtype by @negvet in #2322
• Contrain yapf verison lesser than 0.40.0 by @eunwoosh in #2328
• Fix detection e2e tests by @jaegukhyun in #2327
• Mergeback: Label addtion/deletion 1.2.4 --> 1.4.0 by @sungmanc in #2326
• Bump datumaro up to 1.4.0rc2 by @yunchu in #2332
• Tiling Doc for releases 1.4.0 by @eugene123tw in #2333
• Bump otx version to 1.4.0rc2 by @yunchu in #2341
• Add OTX deploy for visual prompting task by @sungchul2 in #2311
• Bump albumentations version in anomaly requirements by @djdameln in #2350
• Update action detection by @jaegukhyun in #2346
• Fix e2e by @jaegukhyun in #2348
• Update ModelAPI in 1.4 release by @sovrasov in #2347
• Add OTX optimize for visual prompting task by @sungchul2 in #2318
• Update detection docs by @jaegukhyun in #2335
• Add visual prompting documentation by @sungchul2 in #2354
• Remove custom modelapi patch in visual prompting by @sungchul2 in #2359
• Fix graph metric order and label issues by @goodsong81 in #2356
• Update multi-label document and conversion script by @sungmanc in #2358
• Update third party programs by @yunchu in #2365
• Make anomaly task compatible with older albumentations versions by @djdameln in #2363
• Fixing detection saliency map for one class case by @negvet in #2368
• Add e2e test for visual prompting by @sungchul2 in #2360
• Fix e2e by @jaegukhyun in #2366
• Add Dino head unit tests by @jaegukhyun in #2344
• Update for release 1.4.0rc2 by @yunchu in #2370

Full Changelog: 1.4.0rc1...1.4.0rc2

Release 1.2.4

Bug fixes

• Fix 'None' node issue in label schema mapping in case of label deletion (#2300)
• Squeeze anomaly mpas in metadata (#2310)
• Fix semantic segmentation soft prediction dtype (#2321)
• Fix label deletion issue (#2315)

Enhancements

• Per-class saliency maps for M-RCNN (#2301)
• Disable semantic segmentation soft prediction processing (#2302)
• Update export and nncf hyperparameters (#2306)

Full Changelog: 1.2.3.5...1.2.4

Release 1.2.4rc1

What's Changed

• Disable semantic segmentation soft prediction processing by @negvet in #2302
• Per-class saliency maps for M-RCNN by @negvet in #2301
• Fix 'None' node issue in label schema mapping in case of label deletion by @goodsong81 in #2300
• Update export and nncf hyperparameters by @jaegukhyun in #2306
• [FIX] Squeeze anomaly maps in metadata by @djdameln in #2310
• Fix semantic segmentation soft prediction dtype by @negvet in #2321
• added publish workflow by @yunchu in #2325
• Resolve label deletion issue by @sungmanc in #2315

Full Changelog: 1.2.3.5...1.2.4rc1

1.4.0rc1

What's Changed

• [CI] Enable pre-merge tests & doc build on release branch by @goodsong81 in #1795
• [DOC] Added test workflow for testing docs build and publish by @yunchu in #1809
• [DOC] Tutorial for self-supervised learning by @sungchul2 in #1788
• [DOC] New docs workflow for releases by @yunchu in #1810
• [DOC][CI] Updated docs workflows by @yunchu in #1831
• [FEATURE] Add FP16 IR export support by @sovrasov in #1683
• [DOC][CI] Updated docs workflow by @yunchu in #1832
• [HOTFIX][Release1.0] Patches for Geti integration by @goodsong81 in #1836
• [Doc] Apply text feedbacks from OV doc team by @chuneuny-emily in #1838
• [HOTFIX][RELEASE1.0] Fix missing classes in the first incremental round & cleanup classification utils by @sovrasov in #1839
• [DOC] Add classification ref results by @sovrasov in #1837
• [FIX][Release1.0] Fix action task sample codes and unit tests by @jaegukhyun in #1834
• [HOTFIX][RELEASE1.0] Fixed label_scheme mismatch in classification by @harimkang in #1841
• [FIX] Resolve pre-commit issue by @harimkang in #1845
• [ENHANCE] Raise ValueError otx build with wrong training type by @harimkang in #1844
• [ENHANCE] Update backbone lists by @harimkang in #1835
• [DOC][CI] Added workflow for docs clean up by @yunchu in #1840
• [DOC] Proofread readme by @chuneuny-emily in #1847
• [FIX] Align export cli by @cih9088 in #1843
• [FIX] Fix datumaro to otx classification adapter by @sovrasov in #1854
• [CI] Changed .sh file's mode by @yunchu in #1853
• [Enhance] Add in-memory caching in dataloader by @vinnamkim in #1694
• [DOC] Add ref results on multilabel datasets by @sovrasov in #1858
• [Test] Add --test-work-dir option to pytest argparser by @vinnamkim in #1857
• [Enhance] Separate installation for each tasks by @eunwoosh in #1856
• [FIX] Fix label mismatch of evaluation and validation with large dataset in semantic segmentation by @supersoob in #1851
• [Enhance] Run integration test on each task's environment by @eunwoosh in #1859
• [FIX] Enable VOC dataset in OTX by @supersoob in #1862
• [CI][SDL] Added Bandit scanning job to code-scan wf by @yunchu in #1863
• [FIX] hang issue when tracing a stack in certain scenario by @cih9088 in #1868
• [FIX][POT] Set stat_requests_number parameter to 1 by @nikita-savelyevv in #1870
• [Docs] Add in-memory caching guide to quick_start_guide/cli_commands by @vinnamkim in #1860
• [FIX] Match keys of required and options by @sungchul2 in #1864
• [DOC] fix docs by @kprokofi in #1850
• [ENHANCE] Bring mmdeploy to action recognition model export & Test optimization of action tasks by @jaegukhyun in #1848
• [FIX] Training error when batch size is 1 by @cih9088 in #1872
• [CI] Upgrade actions/setup-python version to v4 by @yunchu in #1871
• [FEATURE] Add embedding of inference configuration to IR for classification by @sovrasov in #1842
• [TEST] Weekly Test by @sungmanc in #1689
• [HOTFIX] Update ipas_default.config by @yunchu in #1877
• [ENHANCE] Integration of recipe and model configuration by @harimkang in #1875
• [FIX] Add config.runner.type condition in runner replacing logic by @JihwanEom in #1876
• [FEATURE] Add MoViNet template for action classification by @JihwanEom in #1742
• [FIX] Arrange scale between bbox preds and bbox targets in ATSS by @jaegukhyun in #1880
• [Enhance] Add PR template by @sungmanc in #1881
• [FEATURE] Add train-type parameter to otx train by @sovrasov in #1874
• [FIX] Replace deprecated task_type to task in ConfigManager message by @JihwanEom in #1882
• [FIX] Add SKIP mark and edit eval threshold for E2E tests by @sungmanc in #1878
• [FIX] Modify wrong configs path by @harimkang in #1886
• [TEST] Adding fuzzing test by @yunchu in #1884
• [FIX] Error message is printed after otx train is done by @eunwoosh in #1887
• [ENHANCE] Add skip mark to MoviNet POT, Instance-seg and Adjust export threshold by @sungmanc in #1889
• [FIX] Dataset Adapter: Avoid duplicated annotation and permit empty image by @sungmanc in #1873
• [FIX] Update selfsl args in test_classification.py by @sungchul2 in #1890
• [FIX] Fix the condition of supcon for seg by @sungchul2 in #1883
• [FEATURE] Add Semi-SL multilabel classification algorithm by @sovrasov in #1805
• [FEATURE] Integrate multi-gpu training for semi-supervised learning and self-supervised learning by @supersoob in #1534
• [FIX] Prevent torch 2.0.0 installation by @JihwanEom in #1896
• [CI] Ignore coverage upload failure by @yunchu in #1897
• [ENHANCE] Move detection-related modules from mpa to adapter by @harimkang in #1894
• [FIX] Add skip instance segmentation deploy E2E tests by @harimkang in #1898
• [ENHANCE] Parametrize saliency maps dumping in export (part 2) by @GalyaZalesskaya in #1888
• [RELEASE] Merge back v1.0.1 by @goodsong81 in #1899
• [FIX] Fix typo for backbone list by @harimkang in #1900
• [FIX] Handle unpickable update_progress_callback by @cih9088 in #1892
• [CI] Updated daily workflow by @yunchu in #1904
• [FIX] re-bugfix: ATSS head loss by @eunwoosh in #1907
• [CVS-106621] Change InferenceParameters for demo by @jaegukhyun in #1942
• Change num_iters for e2e object detection test by @jaegukhyun in #1943
• Merge back changes in release 1.1.0 by @goodsong81 in #1947
• MIT->Apache license in PR template by @ashwinvaidya17 in #1949
• Add fp16 export tests by @sovrasov in #1915
• Update regression tests by @sungmanc in #1950
• 🔨 Replace shapes with hazelnut dataset by @ashwinvaidya17 in #1948
• Fix model API task type name classification->Classification by @Wovchena in https://github.com/openvinotoolkit/training_extensio...

Release 1.3.1

Bug fixes

• Fix a bug that auto adapt batch size doesn't work with cls incr case (#2199)
• Fix a bug that persistent worker is True even if num_workers is zero (#2208)

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 1.2.3

Enhancements

• Improve warning message for tiling configurable parameter (#2193)

Bug fixes

• Return raw anomaly map instead of colormap as metadata to prevent applying colormap conversion twice (#2217)
• Hotfix: use 0 confidence threshold when computing best threshold based on F1