Release 2.0.0

What's Changed - Brief Version

NOTES

OpenVINO™ Training Extensions which version 2.0.0 has been updated to include refactoring of the overall architecture and functional updates. Users should install the new environment.

New features

• Enable New design to provide a more seamless API/CLI that delivers the value of OTX: Product Design
• Moved away from MMLab's libraries to provide a Lightning-based core and training pipeline
• Use Lightning-based modules and trainers to deliver APIs/CLIs in a more user-friendly way
• Support Intel devices for accelerating deep learning model training

Enhancements

• Support more models for each task
• Improve the API so user can configure efficient training with shorter code
• Provide more customize settings through the CLI and API
• Enhance the Auto-Configuration feature and made it available in the API

Bug fixes

• Fixing some minor issues

What's Changed - Full Version

• Initial commit by @vinnamkim in #2670
• Initialize classification for the v2 by @sungmanc in #2678
• Enable local package installation and CLI draft implementation by @harimkang in #2679
• Add IterationTimer callback by @vinnamkim in #2682
• Enable Ruff-UP007 (PEP604) by @harimkang in #2684
• Fix coverage error on CI by @vinnamkim in #2687
• Make subset_name configurable by @vinnamkim in #2685
• Fix tox configuration in CI by @harimkang in #2689
• Add memory caching by @vinnamkim in #2686
• Migrate Geti object detection models by @jaegukhyun in #2690
• Update recipe data subsets (hot-fix) by @jaegukhyun in #2694
• Fix warning from batch_size auto-inference failure by @vinnamkim in #2692
• Add Simple Integration test (CLI - 'otx train') by @harimkang in #2695
• Turn on setting seed and deterministic flags by @vinnamkim in #2700
• Change docstring location and learning hyperparams for object detection by @jaegukhyun in #2699
• Add semantic segmentation models to new design by @kprokofi in #2693
• Fix integration test config settings by @harimkang in #2701
• Organize and Enable callbacks per task by @harimkang in #2703
• Mem cache with resize by @vinnamkim in #2702
• Add 'otx test' command by @harimkang in #2707
• Add torchvision v2 transform by @vinnamkim in #2709
• Bring the SOTA(DINOv2) model to the classification task by @sungmanc in #2708
• Add unit tests for otx/algo/detection and otx/core/data/(base, detection) by @jaegukhyun in #2711
• Add Instance Segmentation models to new design by @eugene123tw in #2698
• Add custom model example by @vinnamkim in #2712
• Change the cls head to make them same with the otx1.5 by @sungmanc in #2716
• Add regression test workflow by @vinnamkim in #2718
• Fix otx_train by @vinnamkim in #2720
• Add learning rate monitor by @vinnamkim in #2721
• Add testmetric to the log file by @sungmanc in #2722
• Add unit tests for segmentation by @kprokofi in #2715
• Revise cli unit test by @vinnamkim in #2723
• Fix regression test guide doc by @vinnamkim in #2726
• Fix DINOv2 data transform, add linear probe recipe by @sungmanc in #2725
• Enable ruff-format into v2 by @harimkang in #2735
• Add DinoV2 for semantic segmentation by @kprokofi in #2731
• Enable action classification for v2 by @jaegukhyun in #2737
• Initialize product design doc and add design proposal process to contribution guide by @vinnamkim in #2733
• Enable multilabel classification by @sungmanc in #2739
• Fix MM*DataPreProcessor issue with a more elegant way by @vinnamkim in #2744
• Fix bug by resetting timer every epoch begins by @vinnamkim in #2746
• Update design doc core part by @vinnamkim in #2741
• Enable action detection task for v2 by @jaegukhyun in #2747
• Add GPU marker to the regression test by @vinnamkim in #2754
• Enhance classification task's model performance by @vinnamkim in #2749
• Add detection regression test by @jaegukhyun in #2752
• Patch mmaction2 at installation by @vinnamkim in #2756
• Update PR Reviewers by @harimkang in #2759
• Support TorchVision v2 transforms with ImageInfo by @vinnamkim in #2750
• Add OpenVINO IR inference for CLS, OD, SSEG, IS tasks + OMZ OV models validation support by @kprokofi in #2748
• Enable smart weight loading by @jaegukhyun in #2758
• Enable h-label classification by @sungmanc in #2761
• Add medium and large reg tests for multiclass and multilabel classifications by @sungmanc in #2770
• Update weight copy logic of smart weight loading to support various cases by @jaegukhyun in #2767
• Fix error on logging hparams of OTXDataModule by @vinnamkim in #2773
• Pre-work for the design change: create own python class for every model by @vinnamkim in #2775
• Add asynchronous OV IR inference by @kprokofi in #2769
• Update object detection regression by @jaegukhyun in #2777
• Add gradient clipping and sync hyper parameters wit OTX1.x by @jaegukhyun in #2786
• Add Engine class for OTX Entry Point by @harimkang in #2778
• Update val_check_interval default value in Engine by @harimkang in #2798
• Apply FP16 mixed precision for all tasks by @jaegukhyun in #2789
• Hotfix for segmentation + raising an error when metric is None by @kprokofi in #2788
• Create own python class for every model for action cls, det, and classification tasks by @vinnamkim in #2776
• Add adaptive validation interval by @vinnamkim in #2799
• Fix multilabel training on non-trivial data by @sovrasov in #2796
• Add ImageColorChannel to ImageInfo by @vinnamkim in #2792
• Create own python class for every model inst seg by @vinnamkim in #2781
• Create own python class for every model semantic seg by @vinnamkim in #2782
• Create own python class for every model det by @vinnamkim in #2783
• Semantic segmentation regression tests by @kprokofi in #2809
• Modify Detection's configs by @jaegukhyun in #2811
• Add Instance Segmentation Regression Test by @eugene123tw in #2807
• Et/v2 register explain hook by @vinnamkim in #2810
• Add visual prompting fine tuning task by @sungchul2 in #2780
• Enable performance test (formerly known as a regression test) in CI by @chuneuny-emily in #2803
• Fix Hlabel and add medium test case by @sungmanc in #2804
• Fix hlabel regtest by @sungmanc in #2820
• Fix regression tests by @kprokofi in #2813
• Update Datumaro version to 1.6.0rc1 by @vinnamkim in #2816
• Remove center crop in cls configs by @sovrasov in #2817
• Fix predict checkpoint by @negvet in #2819
• Refactor configuration & CLI by @harimkang in #2805
• Fix cyclic import by @vinnamkim in #2827
• Introduce the logic to load the OTX1.X checkpoints (Det, iSeg, sSeg, Action cls, det) by @sungmanc in #2812
• Update for_developer md files by @harimkang in #2829
• Tiling v2 by @eugene123tw in #2764
• Hot-Fix argument order is different issue in visual_prompting by @harimkang in #2833
• Fix regression test for visual prompting & Add ResizeToLongestEdge by @sungchul2 in #2822
• Add seed_everything into CLI side by @harimkang in #2836
• Added multilabel and hierarchical classification models for IR inference by @kprokofi in #2815
• Accept a list of tvt_v2.Transform or tvt_v2.Compose Python object for data pipeline by @vinnamkim in #2838
• Introduce the logic to load the OTX1.x classification models by @sungmanc in #2828
• Fix to accept non-partial optimizers as well in OTXLitModule by @harimkang in #2840
• Add XAI entry point by @negvet in #2818
• Add ignored label to all task(cls, det, ins-seg, seg) by @jaegukhyun in #2839
• Fix unit tests by @jaegukhyun in #2844
• Hot fix engine.test checkpoint by @harimkang in #2846
• Enable Auto-Configuration by @harimkang in #2841
• Add zero-shot visual prompting task by @sungchul2 in #2834
• Change target ci-runner to gh-hosted runner for unittests by @yunchu in #2847
• Add XAI detection hook by @GalyaZalesskaya in #2842
• Add Engine.from_config by @harimkang in #2852
• Make objects pickleable by @eunwoosh in #2845
• Deprecate python 3.8 and add python 3.11 unit-test by @harimkang in #2854
• Add ViT-B for visual prompting by @sungchul2 in #2850
• Enable VIT ReciproCAM by @negvet in #2849
• Add export entry point by @sovrasov in #2774
• Fix import error when launching a debugger by @vinnamkim in #2862
• Add list_models function to return a list of OTX available models & 'otx find' CLI command by @harimkang in #2856
• Add auto-config num_classes in CLI side by @harimkang in #2861
• Add help verbosity with export & explain by @harimkang in #2863
• Introduce warm-up and change the LRPlateau monitor value, Enable Early stopping by @sungmanc in #2857
• Fix CLI otx predict by @GalyaZalesskaya in #2865
• Fix openxlab version by @harimkang in #2872
• Add data filtering by @jaegukhyun in #2823
• Copy HPO algorithm into OTX 2.0 by @eunwoosh in #2864
• Add PTQ optimization by @sovrasov in #2825
• Hot-Fix conflict between nncf & openxlab by @harimkang in #2875
• Support activation map by @negvet in #2860
• Enable export of DinoV2 models by @sovrasov in #2869
• Transparent forwarding of deterministic parameter to trainer by @sovrasov in #2866
• Update year in PR template by @sovrasov in #2878
• Add exportable code + demo by @kprokofi in #2824
• Add regression test for zero-shot visual prompting by @sungchul2 in #2858
• Add aggregator for LiteHRNet by @kprokofi in #2876
• Add rotated detection by @jaegukhyun in #2873
• Add force fp32 to focal loss forward by @jaegukhyun in #2880
• Add XAI hook for Mask-RCNN by @GalyaZalesskaya in #2870
• Add label_group information to the LabelInfo, Fix label_tree_edges information by @sungmanc in #2883
• Resolve issue where the model configuration was missing in output configuration file by @harimkang in #2886
• Change the setting of regression tests according to the introducing early-stopping by @sungmanc in #2885
• Add more convenient way to override callbacks and loggers per configuration file by @harimkang in #2884
• Tiling Speed & Accuracy Enhancement by @eugene123tw in #2867
• Add mmdeploy export by @eunwoosh in #2806
• Hot-fix Exclude tiling model from export integration test by @eunwoosh in #2895
• Add meta_info update in segmentation dataset class by @harimkang in #2890
• Hot-fix regression test by @harimkang in #2894
• Add dumping of hierarchical config on export by @sovrasov in #2868
• Add --open-subprocess option for integration tests by @vinnamkim in #2896
• Add Auto-Configuration for OVModel by @harimkang in #2897
• Add Dice metric for semantic segmentation by @kprokofi in #2877
• Add workaround for h-cls inference by @sovrasov in #2904
• Update data pipeline for DinoV2 cls by @sovrasov in #2898
• Fix to allow Scheduler to be configured as a list and fix a bug in the AdaptiveTrainingCallback by @harimkang in #2910
• Separate integration tests by task by @harimkang in #2914
• Refine Tile Processing and Expand Tile Configuration by @eugene123tw in #2903
• Migrate SSD anchor generator by @jaegukhyun in #2915
• Add entities for storing explain results by @sovrasov in #2913
• Enable HPO in OTX 2.0 by @eunwoosh in #2912
• Move Linearwarmup scheduler from base to the algo by @sungmanc in #2924
• Fix SegNext data config by @kprokofi in #2928
• Enable auto_num_workers by @eunwoosh in #2931
• Enable forward_explain() for classification by @negvet in #2918
• Upgrade deps by @sovrasov in #2939
• Enable to select the metric at the configuration level by @sungmanc in #2917
• Add OTX CLI Workspace enhancements by @harimkang in #2934
• Add F1 score as object detection, instance segmentation metric by @jaegukhyun in #2962
• Fix a bug where scheduler overrides didn't work correctly by @harimkang in #2967
• Add unannotated_bg_ratio parameter to pre_filtering function by @harimkang in #2965
• Update .latest cache directory with subcommand links by @harimkang in #2969
• Hot-fix wrong integration test case by @harimkang in #2970
• Add missing export parameters for mlc and h-cls by @sovrasov in #2968
• Action cls export by @jaegukhyun in #2961
• Enable forward_explain() for deit by @negvet in #2957
• Update regression test metric of object detection and instance segmentation by @jaegukhyun in #2972
• Add OV Tiler for 2.x by @eugene123tw in #2925
• Reconstruct the metric folder and Refactoring metric related things by @sungmanc in #2964
• Deprecate support for python 3.9 in OTX 2.0 by @harimkang in #2991
• Align Engine.from_config with CLI behavior by @harimkang in #2973
• Enhance ImageInfo.repr() by @vinnamkim in #2992
• ✨ Add Anomalib v1 tasks to OTX v2 by @ashwinvaidya17 in #2902
• Introduce the accuracy and change meta_info -> label_info by @sungmanc in #2994
• Enable sal map postprocessing by @negvet in #2977
• HotFix the seg metric configuration by @sungmanc in #3003
• Ensure backward compatibility for MobileNetv3 at Hlabel classification by @sungmanc in #3002
• Implement sampler API & add BalancedSampler by @harimkang in #2993
• Add test metric info transfer pipeline by @jaegukhyun in #2996
• Change CODEOWNERS list by @harimkang in #3008
• Add tests: forward explain + predict with explain by @negvet in #2998
• Improve inst seg perf by @vinnamkim in #3005
• Fix tile bug & add more tests by @eugene123tw in #3012
• Add perf benchmark tests for v2 by @goodsong81 in #3004
• Set num_sanity_val_steps to 0 as default in OTX CLI by @harimkang in #3010
• Enable Balanced-Sampler into cls, det, ins-seg by @harimkang in #3019
• Add Class-Incremental Sampler by @harimkang in #3018
• Add auto-config default recipe for H-label classification by @harimkang in #3031
• Support dumping saliency maps by @negvet in #3011
• Enable export for visual prompting models by @sungchul2 in #2955
• Add checkpoint argument to optimize() by @sovrasov in #3028
• Add tile regression tests and tile recipe by @eugene123tw in #3037
• Support explain for OV models by @GalyaZalesskaya in #3027
• Add docker image build script by @vinnamkim in #3036
• Hot fix for export inference test by @jaegukhyun in #3045
• Add Action classification regression test by @jaegukhyun in #3041
• Update pre_merge workflow to upload coverage to codecov by @yunchu in #3043
• Update code-scan workflow by @yunchu in #3044
• Support feature vector for cls by @negvet in #3047
• Fix warmup scheduler and add patience update for adaptive interval by @sungmanc in #3056
• Add torchvision models for multi-class classification by @harimkang in #3034
• Training performance (speed) review for the semantic seg tasks by @vinnamkim in #3051
• Add RTMDet Inst Tiny by @eugene123tw in #3039
• Add optimize for visual prompting to 2.0 by @sungchul2 in #3040
• Add hlabel parent resolver and integrate HLabelData + HLabelInfo --> HLabelInfo by @sungmanc in #3046
• Refine perf benchmark v2 by @goodsong81 in #3052
• Dedicate perf-benchmark v2 CI host by @goodsong81 in #3063
• Make the same config with OTX1.x : Edit the default factor for lr scheduler and update warmup_step logic by @sungmanc in #3060
• Adapt mask prediction to rotated box prediction by @eugene123tw in #3029
• Add dependabot.yaml by @yunchu in #3053
• Update for resolving security issues by @yunchu in #3055
• Fix trivy-scan testenv by @yunchu in #3072
• Add Movinet to action classification by @jaegukhyun in #3059
• Add classification backbones unit test by @jaegukhyun in #3065
• Add test models for perf benchmark v2 by @goodsong81 in #3083
• Fix the Anomaly Task to work & Add integration test for Anomaly by @harimkang in #3007
• Add docs workflow for latest & testing by @yunchu in #3089
• Bring docs from OTX 1.6 to 2.0 by @harimkang in #3087
• Detect num_classes information for the H-label classification by @sungmanc in #3064
• Forward_explain & Export with XAI: Detection & InstSeg by @GalyaZalesskaya in #3079
• Update publish workflow and add publish_internal workflow by @yunchu in #3097
• Add missing workflows on v2 by @yunchu in #3098
• Bump mlflow to 2.11.1 by @yunchu in #3131
• Refactoring for zero-shot standalone usecase by @sungchul2 in #3132
• Change Hlabel Dataset: dummy -> dummy trump card by @sungmanc in #3128
• Add from_model_name in Engine by @harimkang in #3129
• Add unit tests under otx.algo by @jaegukhyun in #3127
• Support perf result reference check for benchmark v2 by @goodsong81 in #3140
• Remove unused files & Add coverage.omit init and recipes by @harimkang in #3143
• Add changelog and update readme by @yunchu in #3130
• Add automatic docs/source builder to check the output easily by @sungmanc in #3138
• Update Anomaly docs by @ashwinvaidya17 in #3151
• Add DRAEM anomaly model by @ashwinvaidya17 in #3152
• Add more backbone list for tv & fix exported_params by @harimkang in #3144
• Bump sphinx from 5.3.0 to 7.2.6 by @dependabot in #3134
• Add Anomaly OpenVINO model by @ashwinvaidya17 in #3146
• Fix wrong ov datamodule update in predict, test, optimize by @harimkang in #3139
• Add classification tutorial by @sungmanc in #3153
• Promote OTXModel to PyTorchLightningModule and deprecate OTXLitModule by @vinnamkim in #3155
• Update trivy scanning to use actions by @yunchu in #3169
• Implement unit test for mmdeploy exporter for develop branch by @eunwoosh in #3184
• Change code using fixed subset name in HPO by @eunwoosh in #3101
• Refactor label info by @vinnamkim in #3180
• Bump docstring-parser from 0.15 to 0.16 by @dependabot in #3149
• Bump jsonargparse from 4.27.1 to 4.27.6 by @dependabot in #3148
• Fix F1 metric by @vinnamkim in #3187
• Update visual prompting refactoring to develop by @sungchul2 in #3193
• Bring RTMDetInst Optimization and tile unit test to Dev by @eugene123tw in #3195
• Add unit test for export for develop branch by @eunwoosh in #3212
• Set model label info first by @jaegukhyun in #3213
• Refactor optimizer and lr scheduler part by @vinnamkim in #3216
• Refactor XAI data entities by @vinnamkim in #3230
• Bump sphinx-copybutton from 0.5.0 to 0.5.2 by @dependabot in #3199
• Bump importlib-resources from 6.1.3 to 6.4.0 by @dependabot in #3196
• Migrate docs/ by @harimkang in #3243
• Push unit tests and some bug fixes to develop by @harimkang in #3224
• Migrate from releases/2.0.0 by @jaegukhyun in #3245
• Add Configuration Converter in otx.tools by @harimkang in #3254
• Fix tile PTQ transform by @eugene123tw in #3261
• Refactor XAI model part - classification tasks by @vinnamkim in #3242
• Fix tile CLI by @eugene123tw in #3204
• Bump pillow from 10.2.0 to 10.3.0 in /for_developers/regression_test by @dependabot in #3263
• Enable e2e for develop by @yunchu in #3260
• Fix a bug occurred when training a RTMdet tiny model (dev branch) by @eugene123tw in #3272
• Fix semantic seg ground truth mask extraction by @vinnamkim in #3270
• Merge pytest parser options by @vinnamkim in #3269
• Update ci-runner start script to use host shm size as-is by @yunchu in #3274
• Fix Dice score. Add mIoU metric for semantic segmentation by @kprokofi in #3264
• Refactor to implant optimizer and scheduler into model code by @vinnamkim in #3258
• Fix tile polygon transform function (Dev branch) by @eugene123tw in #3288
• Update publish workflow to add contents write permission by @yunchu in #3286
• Temporary fix for jsonargparse issue on Python3.11.9 by @vinnamkim in #3291
• Extend exportable code by @kprokofi in #3284
• Migrate SSD from mmdet to OTX by @jaegukhyun in #3279
• Bump aquasecurity/trivy-action from 0.18.0 to 0.19.0 by @dependabot in #3250
• Bump onnx from 1.15.0 to 1.16.0 by @dependabot in #3246
• Bump oss2 from 2.17.0 to 2.18.4 by @dependabot in #3248
• Bump pre-commit from 2.20.0 to 3.7.0 by @dependabot in #3249
• Refactor export part 1 - export parameters by @vinnamkim in #3289
• Replace mmcv nms to torchvision nms by @jaegukhyun in #3300
• Decoupling mmdet structures Part 1. by @jaegukhyun in #3301
• v2.1: XAI clean-up & XAI unit tests by @GalyaZalesskaya in #3303
• Refactor checkpiont logic by @vinnamkim in #3302
• Update readme, changelog, release note and version string by @yunchu in #3320
• XAI tor tiling: detection, instance segmentation by @GalyaZalesskaya in #3297
• Fix e2e issues by @yunchu in #3313
• Decoupling mmdet structures part2 by @jaegukhyun in #3315
• Bump actions/setup-python from 5.0.0 to 5.1.0 by @dependabot in #3251
• Bump sqlparse from 0.4.4 to 0.5.0 in /for_developers/regression_test by @dependabot in #3328
• Bump idna from 3.6 to 3.7 in /for_developers/regression_test by @dependabot in #3306
• Move XPU/Opt related PRs from releases2.0 to develop by @kprokofi in #3295
• HPO code flow modification by @eunwoosh in #3259
• Refactor tiling data typing by @vinnamkim in #3331
• Fix flaky test: TestCustomRTMDetInsSepBNHead.test_mask_pred by @eugene123tw in #3333
• Decouple mmpretrain: Mobilenet-V3-large by @harimkang in #3271
• Support benchmark history summary v2.1 by @goodsong81 in #3312
• Update IPEX version in installation guide documentation by @eunwoosh in #3342
• Refactor model export part 2: Add a dedicated forward function for model export by @vinnamkim in #3317
• Add SchedulerCallableSupportHPO by @vinnamkim in #3334
• Update CODEOWNERS by @harimkang in #3341
• Pin artifact download action on perf-benchmark wf by @yunchu in #3348
• Decouple detection data pipeline from mmdet by @sungchul2 in #3253
• 🗑️ Remove Draem by @ashwinvaidya17 in #3345
• Fix spatial resolution for saliency_maps in torchvision_models by @GalyaZalesskaya in #3336
• Re-org deps for ci and pin deps for benchmark workflow by @yunchu in #3349
• Revisit LabelInfo to make it immutable on OTXModel side as possible by @vinnamkim in #3344
• Add RandomResizedCrop for cv2 & Change mv3 pipelines by @harimkang in #3352
• Reduce default batch size of action detection by @jaegukhyun in #3367
• Bring ConvModule from mmcv.cnn.bricks by @harimkang in #3368
• Fix jsonargparse cannot parse lambda default arguments by @vinnamkim in #3370
• Revisit TileConfig propagation to make it immutable on OTXModel side as possible by @vinnamkim in #3372
• Fix reference info path for zero-shot learning by @sungchul2 in #3354
• Decouple mmpretrain: Efficientnet-V2-S by @harimkang in #3353
• XAI support for rmdet_inst_tiny by @GalyaZalesskaya in #3356
• Enable SSD export through NativeExporter by @jaegukhyun in #3366
• Fix incorrect cache loading by @harimkang in #3384
• Add adaptive batch size feature by @eunwoosh in #3378
• Decouple mmpretrain: Efficientnet-B0 by @harimkang in #3380
• Fix explain comment in e2e cli test by @jaegukhyun in #3391
• MMDet MaskRCNN ResNet50/SwinTransformer Decouple by @eugene123tw in #3281
• Decouple YOLOX : model migration by @sungchul2 in #3382
• Update PR template by @yunchu in #3393
• Bump mlflow from 2.11.1 to 2.12.1 in /for_developers/regression_test by @dependabot in #3364
• Bump actions/upload-artifact from 4.3.1 to 4.3.2 by @dependabot in #3361
• Enable parameter of scheduler for HPO candidate by @eunwoosh in #3390
• Migrate ATSS from mmdet by @jaegukhyun in #3386
• Decouple YOLOX : unit tests for model migration by @sungchul2 in #3398
• Decouple semantic segmentation models by @kprokofi in #3376
• Decouple mmpretrain: DeiT-Tiny by @harimkang in #3397
• Update to datumaro==1.6.1rc3 by @itrushkin in #3377
• Decouple mmengine: Bring mmengine.BaseModule by @harimkang in #3402
• Decouple mmpretrain: Head, Loss by @harimkang in #3404
• Upgrade OV, NNCF and MAPI by @sovrasov in #3335
• Update e2e test scope by @chuneuny-emily in #3403
• Add result csv upload job for daily workflow by @yunchu in #3399
• Downgrade OV to 2024.0 by @sovrasov in #3410
• Fix e2e tests by @kprokofi in #3411
• Deprecate mm compatible detectors by @jaegukhyun in #3405
• Add unit test for engine/hpo by @eunwoosh in #3406
• Decouple YOLOX : export by @sungchul2 in #3407
• Align Dino-v2 with other models by @harimkang in #3409
• Refactor anomaly dataset and data module by @wonjuleee in #3414
• Remove mmengine from semantic segmentation task by @kprokofi in #3424
• Update HPO to support XPU by @eunwoosh in #3417
• Refine classification OV export by @sovrasov in #3408
• Change location of stack_batch by @jaegukhyun in #3413
• Update to use BaseModule on detection by @sungchul2 in #3415
• Fix Benchmark Test by @harimkang in #3426
• Bump idna from 3.6 to 3.7 in /.ci/requirements/publish by @dependabot in #3363
• Update version string and changelog by @yunchu in #3427
• Remove mm related configs by @jaegukhyun in #3419
• Fix .dockerignore to copy .ci directory by @wonjuleee in #3428
• Remove cuda version focal loss by @jaegukhyun in #3431
• Remove left mmengine things in object detection by @jaegukhyun in #3432
• Make NP TV conversion configurable by @harimkang in #3429
• MaskRCNN Native Exporter by @eugene123tw in #3412
• Fix data pipeline by @sungchul2 in #3418
• Implement unit test for adaptive batch size by @eunwoosh in #3430
• Adjust multi-label classification parameters by @harimkang in #3436
• Remove code to patch nms and roi_align in mmXX for XPU by @eunwoosh in #3434
• Change default fmeasure confidence_threshold from 0 to None by @jaegukhyun in #3441
• Modify label info comparison by @yunchu in #3442
• Update segmentation documentation by @kprokofi in #3425
• Generate label info from v1 checkpoint by @jaegukhyun in #3444
• Add polygons support for semantic segmentation by @kprokofi in #3439
• Add unit tests for detectors' forward function by @jaegukhyun in #3447
• MaskRCNN backword compatibility by @eugene123tw in #3451
• Decouple BaseModel by @eugene123tw in #3450
• Use parent dataset indexing for Datumaro DatasetSubset by @itrushkin in #3438
• Fix Classification v1 checkpoint compatibility by @harimkang in #3458
• [Release] Change load_stat_dict to on_load_checkpoint by @jaegukhyun in #3457
• Consider loop type during adaptive batch size by @eunwoosh in #3461
• Fix Score Computation by @ashwinvaidya17 in #3452
• Fix mixed training and validation subset items by @wonjuleee in #3460
• Fix e2e xai tests by @negvet in #3462
• Revert mrcnn for test_otx_explain_e2e_cli by @negvet in #3466
• Add unit-test for otx.algo.modules by @harimkang in #3465
• Update XPU plugin unit test by @eunwoosh in #3464
• Add unit-test for classification by @harimkang in #3463
• Add additional otx algo det unit tests by @jaegukhyun in #3467
• Add unit-test for mmengine & mmcv util functions by @harimkang in #3468
• Add unit tests for algo/segmentation by @eunwoosh in #3469
• MMDet RTMDet Inst decoupling by @eugene123tw in #3433
• Add metrics to Anomaly OV Task by @ashwinvaidya17 in #3471
• Exclude exportable_code from unit test coverage by @eunwoosh in #3473
• Revert scheduling parameters for multi-label-cls model by @harimkang in #3474
• Enable instance segmentation data pipeline by @sungchul2 in #3435
• Fix anomaly export _get_values_from_transforms function by @harimkang in #3477
• Update to use tiling dataset for instance tiling testing by @chuneuny-emily in #3475
• Raise an error if dino is moved to XPU by @eunwoosh in #3480
• Update docs & dependency and remove task dependency in otx.algo by @harimkang in #3476
• Remove weight initialization arguments for ATSS by @wonjuleee in #3483
• Fix VP test issue by @chuneuny-emily in #3482
• Remove pad_shape in ImageInfo by @sungchul2 in #3479
• Update benchmark by @chuneuny-emily in #3486
• Set stfpm precision to 32 by @ashwinvaidya17 in #3484
• Change OV model scales by @ashwinvaidya17 in #3485
• Decouple mmdet inst seg data pipeline by @eugene123tw in #3472
• Remove mmengine imports by @harimkang in #3487
• Remove left mmdet dependency by @jaegukhyun in #3488
• 🐞 Set metric threshold by @ashwinvaidya17 in #3498
• Update image_decode_context flow with RGB by @harimkang in #3501
• Update for security issues by @yunchu in #3500
• Fix maskrcnn accuracy drop on tiling by @eugene123tw in #3502
• Update dependencies for regression-test in for_developers folder by @yunchu in #3508
• Fix e2e xai instance segmentation tests by @GalyaZalesskaya in #3505
• Fix ATSS recipes by @wonjuleee in #3511
• Remove LITMODULE_PER_TASK is No Longer Used by @harimkang in #3512
• Disable Resnext101_ATSS model on XPU by @eunwoosh in #3514
• Remove invalid perf benchmark reference history for v2.0.0 by @yunchu in #3517
• Fix a bug that dino_v2 model can't be run w/ HPO by @eunwoosh in #3518
• Fix detection export performance degradation by @sungchul2 in #3520
• Revert metrics threshold by @ashwinvaidya17 in #3528
• Tile with full img optional by @eugene123tw in #3530
• Add a feature to adapt max value of HPO batch size search space by @eunwoosh in #3532
• Remove duplicates in get_idx_list_per_classes by @harimkang in #3537
• Revisit Docker image build script by @vinnamkim in #3536
• Bump datumaro version to 1.6.1 by @yunchu in #3535
• Add F1 metric computation during detection tasks by @wonjuleee in #3539
• Fix yolox export perf degradation by @sungchul2 in #3534
• Fix MaskRCNN IR Accuracy Drop by @eugene123tw in #3540
• Hotfix/geti integration by @vinnamkim in #3543
• Fix Anomaly OV export flag by @ashwinvaidya17 in #3558
• Fix accuracy drop in MaskRCNN by @eugene123tw in #3562
• Update codeql workflow to generate a report by @yunchu in #3559
• Fix & Refine HPO by @eunwoosh in #3565
• Fix Optimize in Anomaly Task by @ashwinvaidya17 in #3561
• Update dependencies by @yunchu in #3570
• Bump anomalib from 1.0.1 to 1.1.0 by @harimkang in #3572
• Fix F1 instance seg accuracy drop by @eugene123tw in #3578
• Update for using locally stored pretrained weights in the CI for torch hub models by @yunchu in #3541
• Update ATSS input size to 992x800 by @wonjuleee in #3588
• Update 2.0.0 CHANGELOG by @harimkang in #3589
• Update for release 2.0.0 by @yunchu in #3594
• Update publish workflow for pure python pkg by @yunchu in #3603
• Fix setuptools dependency by @harimkang in #3609

New Contributors

• @itrushkin made their first contribution in #3377

Full Changelog: 1.6.2...2.0.0