merge releases2.2.2

.github/workflows/pre_merge.yaml

@@ -153,3 +153,32 @@ jobs:
           rm /tmp/requirements.txt
       - name: Run Integration Test
         run: tox -vv -e integration-test-${{ matrix.task }}
+  Integration-Test-Large:
+    if: |
+      github.event.pull_request.draft == false &&
+      !(startsWith(github.event.pull_request.title, '[WIP]'))
+    runs-on: [self-hosted, linux, x64, dev, dmount]
+    needs: Unit-Test
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - task: "detection"
+          - task: "instance_segmentation"
+          - task: "semantic_segmentation"
+    name: Integration-Test-Large-${{ matrix.task }}-py310
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
+      - name: Install Python
+        uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # v5.1.1
+        with:
+          python-version: "3.10"
+      - name: Install tox
+        run: |
+          python -m pip install --require-hashes --no-deps -r .ci/requirements.txt
+          pip-compile --generate-hashes --output-file=/tmp/requirements.txt --extra=ci_tox pyproject.toml
+          python -m pip install --require-hashes --no-deps -r /tmp/requirements.txt
+          rm /tmp/requirements.txt
+      - name: Run Integration Test
+        run: tox -vv -e integration-test-${{ matrix.task }}

CHANGELOG.md

@@ -12,6 +12,16 @@ All notable changes to this project will be documented in this file.
   (<https://github.com/openvinotoolkit/training_extensions/pull/4066>)
 - Add GPU memory monitor hook
   (<https://github.com/openvinotoolkit/training_extensions/pull/4118>)
+- Add YOLOv9 model for Object Detection
+  (<https://github.com/openvinotoolkit/training_extensions/pull/3917>, <https://github.com/openvinotoolkit/training_extensions/pull/4026>)
+- Add OV inference for keypoint detection
+  (<https://github.com/openvinotoolkit/training_extensions/pull/3970>)
+- Add tiling for semantic segmentation
+  (<https://github.com/openvinotoolkit/training_extensions/pull/3954>)
+- Add 3D Object Detection task with MonoDETR model
+  (<https://github.com/openvinotoolkit/training_extensions/pull/3979>)
+- Add OpenVINO inference for 3D Object Detection task
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4017>)
 
 ### Enhancements
 
@@ -29,29 +39,6 @@ All notable changes to this project will be documented in this file.
   (<https://github.com/openvinotoolkit/training_extensions/pull/4123>)
 - Improve FMetric computation
   (<https://github.com/openvinotoolkit/training_extensions/pull/4130>)
-
-### Bug fixes
-
-- Fix MaskDINO, MonoDETR recipes
-  (<https://github.com/openvinotoolkit/training_extensions/pull/4080>)
-
-## \[2.3.0\]
-
-### New features
-
-- Add YOLOv9 model for Object Detection
-  (<https://github.com/openvinotoolkit/training_extensions/pull/3917>, <https://github.com/openvinotoolkit/training_extensions/pull/4026>)
-- Add OV inference for keypoint detection
-  (<https://github.com/openvinotoolkit/training_extensions/pull/3970>)
-- Add tiling for semantic segmentation
-  (<https://github.com/openvinotoolkit/training_extensions/pull/3954>)
-- Add 3D Object Detection task with MonoDETR model
-  (<https://github.com/openvinotoolkit/training_extensions/pull/3979>)
-- Add OpenVINO inference for 3D Object Detection task
-  (<https://github.com/openvinotoolkit/training_extensions/pull/4017>)
-
-### Enhancements
-
 - Upgrade OV, MAPI, and NNCF dependencies
   (<https://github.com/openvinotoolkit/training_extensions/pull/3967>)
 - Instance Segmentation Model refactoring
@@ -63,13 +50,44 @@ All notable changes to this project will be documented in this file.
 
 ### Bug fixes
 
+- Fix MaskDINO, MonoDETR recipes
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4080>)
 - Fix a wrong HPO log
   (<https://github.com/openvinotoolkit/training_extensions/pull/3972>)
 - Update model name in rotated detection recipes
   (<https://github.com/openvinotoolkit/training_extensions/pull/4028>)
 - Fix SupCon flag
   (https://github.com/openvinotoolkit/training_extensions/pull/4076)
 
+## \[2.2.2\]
+
+### Bug fixes
+
+- BC improvement
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4154>)
+
+## \[2.2.1\]
+
+### Bug fixes
+
+- Fix empty annotation in tiling
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4124>)
+- Fix patching early stopping in tools/converter.py, update headers in templates, change training schedule for classification
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4131>)
+- Fix tensor type compatibility in dynamic soft label assigner and RTMDet head
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4140>)
+- Fix DETR target class indices are of type long in loss calculations
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4143>)
+- Fix arrow format reader for multiclass ROI case
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4145>)
+
+### Enhancements
+
+- Decouple DinoV2 for semantic segmentation task
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4136>)
+- Update Label Info handling
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4127>)
+
 ## \[2.2.0\]
 
 ### New features
@@ -194,6 +212,8 @@ All notable changes to this project will be documented in this file.
   (<https://github.com/openvinotoolkit/training_extensions/pull/4105>)
 - Disable tiling classifier toggle in configurable parameters
   (<https://github.com/openvinotoolkit/training_extensions/pull/4107>)
+- Fix Ellipse shapes for Instance Segmentation
+  (<https://github.com/openvinotoolkit/training_extensions/pull/4152>)
 
 ## \[v2.1.0\]
 

docs/source/guide/release_notes/index.rst

@@ -4,6 +4,35 @@ Releases
 .. toctree::
   :maxdepth: 1
 
+v2.2.2 (2024.12)
+----------------
+
+Enhancements
+^^^^^^^^^^^^
+
+Bug fixes
+^^^^^^^^^
+
+- BC improvement
+
+v2.2.1 (2024.12)
+----------------
+
+Enhancements
+^^^^^^^^^^^^
+
+- Decouple DinoV2 for semantic segmentation task
+- Update Label Info handling
+
+Bug fixes
+^^^^^^^^^
+
+- Fix empty annotation in tiling
+- Fix patching early stopping in tools/converter.py, update headers in templates, change training schedule for classification
+- Fix tensor type compatibility in dynamic soft label assigner and RTMDet head
+- Fix DETR target class indices are of type long in loss calculations
+- Fix arrow format reader for multiclass ROI case
+
 v2.2.0 (2024.10)
 ----------------
 
@@ -22,6 +51,8 @@ New features
 - Add Semi-SL MeanTeacher algorithm for Semantic Segmentation
 - Update head and h-label format for hierarchical label classification
 - Support configurable input size
+- Revert the old workaround for detection confidence threshold
+- Add Keypoint Detection legacy template
 
 Enhancements
 ^^^^^^^^^^^^
@@ -40,21 +71,38 @@ Enhancements
 - Change sematic segmentation to consider bbox only annotations
 - Relieve memory usage criteria on batch size 2 during adaptive batch size
 - Remove background label from RT Info for segmentation task
+- Enable export of the feature vectors for semantic segmentation task
 - Prevent using too low confidence thresholds in detection
+- Update HPO interface
+- Bump onnx to 1.17.0 to omit CVE-2024-5187
 
 Bug fixes
 ^^^^^^^^^
 
+- Update anomaly base transforms to use square resizing
 - Fix Combined Dataloader & unlabeled warmup loss in Semi-SL
 - Revert #3579 to fix issues with replacing coco_instance with a different format in some dataset
 - Add num_devices in Engine for multi-gpu training
 - Add missing tile recipes and various tile recipe changes
 - Change categories mapping logic
 - Fix config converter for tiling
+- Fix `BboxOverlaps2D` handling of empty ground-truth annotations in datasets.
 - Fix num_trials calculation on dataset length less than num_class
 - Fix out_features in HierarchicalCBAMClsHead
 - Fix multilabel_accuracy of MixedHLabelAccuracy
 - Fix wrong indices setting in HLabelInfo
+- Add legacy template LiteHRNet_18 template
+- Model templates: rename model_status value 'DISCONTINUED' to 'OBSOLETE'
+- Update MRCNN model export to include feature vector and saliency map
+- Upgrade MAPI in 2.2
+- Fix applying model's hparams when loading model from checkpoint
+- Fix incorrect all_groups order configuration in HLabelInfo
+- Fix RTDETR recipes
+- Fix wrong model name in converter & template
+- Fix RTMDet Inst Explain Mode
+- Fix RTDETR Explain Mode
+- Fix classification and semantic segmentation tasks, when ROI provided for images
+- Disable tiling classifier toggle in configurable parameters
 
 v2.1.0 (2024.07)
 ----------------

src/otx/algo/callbacks/adaptive_early_stopping.py

@@ -20,7 +20,7 @@ def __init__(
         self,
         monitor: str,
         min_delta: float = 0.0,
-        patience: int = 3,
+        patience: int = 10,
         verbose: bool = False,
         mode: str = "min",
         strict: bool = True,