Merge branch 'release/v2.0.0' into ci/add-new-gh-actions

Author: samet-akcay

CHANGELOG.md

@@ -8,13 +8,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ### Added
 
-- Add `AUPIMO` tutorials notebooks in https://github.com/openvinotoolkit/anomalib/pull/2330 and https://github.com/openvinotoolkit/anomalib/pull/2336
-- Add `AUPIMO` metric by [jpcbertoldo](https://github.com/jpcbertoldo) in https://github.com/openvinotoolkit/anomalib/pull/1726 and refactored by [ashwinvaidya17](https://github.com/ashwinvaidya17) in https://github.com/openvinotoolkit/anomalib/pull/2329
-
 ### Removed
 
-- Remove `RKDE` in https://github.com/openvinotoolkit/anomalib/pull/2455
-
 ### Changed
 
 ### Deprecated
@@ -23,6 +18,47 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ### New Contributors
 
+## v2.0.0
+
+### Added
+
+- 🚀 Add `Dataclasses` and `PostProcessor` by @djdameln in https://github.com/openvinotoolkit/anomalib/pull/2098
+- 🚀 Add dataclass validators by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2307
+- 🚀 Add Customisable Image Visualizer by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2334
+- 🚀 Metrics redesign by @djdameln in https://github.com/openvinotoolkit/anomalib/pull/2326
+- 🚀 Add `PreProcessor` to `AnomalyModule` by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2358
+- 🚀 Add Multi-GPU Training Support by @ashwinvaidya17 in https://github.com/openvinotoolkit/anomalib/pull/2435
+- 🔨 Refactor: Add missing auxiliary attributes to `AnomalibModule` by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2460
+- 🔨 Rename `AnomalyModule` to `AnomalibModule` by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2423
+
+- 🚀 Add `AUPIMO` metric by [jpcbertoldo](https://github.com/jpcbertoldo) in https://github.com/openvinotoolkit/anomalib/pull/1726 and refactored by [ashwinvaidya17](https://github.com/ashwinvaidya17) in https://github.com/openvinotoolkit/anomalib/pull/2329
+- 📚 Add `AUPIMO` tutorials notebooks in https://github.com/openvinotoolkit/anomalib/pull/2330 and https://github.com/openvinotoolkit/anomalib/pull/2336
+
+### Removed
+
+- 🗑️ Remove RKDE by @ashwinvaidya17 in https://github.com/openvinotoolkit/anomalib/pull/2455
+- 🗑️ Remove rich methods by @ashwinvaidya17 in https://github.com/openvinotoolkit/anomalib/pull/2283
+- 🔨 Replace `imgaug` with Native PyTorch Transforms by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2436
+- 🗑️ Remove task type by @djdameln in https://github.com/openvinotoolkit/anomalib/pull/2450
+
+### Changed
+
+- Refactor Lightning's `trainer.model` to `trainer.lightning_module` by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2255
+- Update open-clip-torch requirement from <2.26.1,>=2.23.0 to >=2.23.0,<2.26.2 by @dependabot in https://github.com/openvinotoolkit/anomalib/pull/2189
+- Update sphinx requirement by @dependabot in https://github.com/openvinotoolkit/anomalib/pull/2235
+- Update ruff configuration by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2269
+- Revert "Update open-clip-torch requirement from <2.26.1,>=2.23.0 to >=2.23.0,<2.26.2" by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2270
+- 🔨 Lint: U\* 🔨 Refactor BaseThreshold to Threshold by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2278
+- 🔨 Enable Ruff Rules: PLW1514 and PLR6201 by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2284
+- 🔨 Update nncf export by @ashwinvaidya17 in https://github.com/openvinotoolkit/anomalib/pull/2286
+- 🔨 Linting: Enable `PLR6301`, # could be a function, class method or static method by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2288
+- 🔨 Restructure unit tests and fix ruff issues by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2306pdate Ruff Config - Add Missing Copyright Headers by @samet-akcay in https://github.com/openvinotoolkit/anomalib/pull/2281
+- 🔨 optimization/quantization added into 500 series by @paularamo in https://github.com/openvinotoolkit/anomalib/pull/2197
+
+### Fixed
+
+- 🐞Replace package_available with module_available by @harimkang in https://github.com/openvinotoolkit/anomalib/pull/2407
+
 ## [v1.2.0]
 
 ### Added

src/anomalib/engine/engine.py

@@ -22,7 +22,6 @@
 from anomalib.deploy import CompressionType, ExportType
 from anomalib.models import AnomalibModule
 from anomalib.utils.path import create_versioned_dir
-from anomalib.visualization import ImageVisualizer
 
 logger = logging.getLogger(__name__)
 
@@ -258,13 +257,13 @@ def _setup_trainer(self, model: AnomalibModule) -> None:
             self._cache.update(model)
 
         # Setup anomalib callbacks to be used with the trainer
-        self._setup_anomalib_callbacks(model)
+        self._setup_anomalib_callbacks()
 
         # Instantiate the trainer if it is not already instantiated
         if self._trainer is None:
             self._trainer = Trainer(**self._cache.args)
 
-    def _setup_anomalib_callbacks(self, model: AnomalibModule) -> None:
+    def _setup_anomalib_callbacks(self) -> None:
         """Set up callbacks for the trainer."""
         _callbacks: list[Callback] = []
 
@@ -279,18 +278,6 @@ def _setup_anomalib_callbacks(self, model: AnomalibModule) -> None:
                 ),
             )
 
-        # Add the post-processor callback.
-        if isinstance(model.post_processor, Callback):
-            _callbacks.append(model.post_processor)
-
-        # Add the metrics callback.
-        if isinstance(model.evaluator, Callback):
-            _callbacks.append(model.evaluator)
-
-        # Add the image visualizer callback if it is passed by the user.
-        if not any(isinstance(callback, ImageVisualizer) for callback in self._cache.args["callbacks"]):
-            _callbacks.append(ImageVisualizer())
-
         _callbacks.append(TimerCallback())
 
         # Combine the callbacks, and update the trainer callbacks.

src/anomalib/models/components/base/__init__.py

@@ -3,7 +3,7 @@
 # Copyright (C) 2022-2024 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 
-from .anomaly_module import AnomalibModule
+from .anomalib_module import AnomalibModule
 from .buffer_list import BufferListMixin
 from .dynamic_buffer import DynamicBufferMixin
 from .memory_bank_module import MemoryBankMixin

…models/components/base/anomaly_module.py …odels/components/base/anomalib_module.pysrc/anomalib/models/components/base/anomaly_module.py renamed to src/anomalib/models/components/base/anomalib_module.py src/anomalib/models/components/base/anomaly_module.py renamed to src/anomalib/models/components/base/anomalib_module.py

@@ -25,6 +25,7 @@
 from anomalib.metrics.threshold import Threshold
 from anomalib.post_processing import OneClassPostProcessor, PostProcessor
 from anomalib.pre_processing import PreProcessor
+from anomalib.visualization import ImageVisualizer, Visualizer
 
 from .export_mixin import ExportMixin
 
@@ -40,8 +41,9 @@ class AnomalibModule(ExportMixin, pl.LightningModule, ABC):
     def __init__(
         self,
         pre_processor: PreProcessor | bool = True,
-        post_processor: PostProcessor | None = None,
+        post_processor: PostProcessor | bool = True,
         evaluator: Evaluator | bool = True,
+        visualizer: Visualizer | bool = True,
     ) -> None:
         super().__init__()
         logger.info("Initializing %s model.", self.__class__.__name__)
@@ -52,11 +54,12 @@ def __init__(
         self.callbacks: list[Callback]
 
         self.pre_processor = self._resolve_pre_processor(pre_processor)
-        self.post_processor = post_processor or self.default_post_processor()
+        self.post_processor = self._resolve_post_processor(post_processor)
         self.evaluator = self._resolve_evaluator(evaluator)
+        self.visualizer = self._resolve_visualizer(visualizer)
 
         self._input_size: tuple[int, int] | None = None
-        self._is_setup = False  # flag to track if setup has been called from the trainer
+        self._is_setup = False
 
     @property
     def name(self) -> str:
@@ -79,28 +82,20 @@ def _setup(self) -> None:
         initialization.
         """
 
-    def _resolve_pre_processor(self, pre_processor: PreProcessor | bool) -> PreProcessor | None:
-        """Resolve and validate which pre-processor to use..
-
-        Args:
-            pre_processor: Pre-processor configuration
-                - True -> use default pre-processor
-                - False -> no pre-processor
-                - PreProcessor -> use the provided pre-processor
+    def configure_callbacks(self) -> Sequence[Callback] | Callback:
+        """Configure default callbacks for AnomalibModule.
 
         Returns:
-            Configured pre-processor
+            List of callbacks that includes the pre-processor, post-processor, evaluator,
+            and visualizer if they are available and inherit from Callback.
         """
-        if isinstance(pre_processor, PreProcessor):
-            return pre_processor
-        if isinstance(pre_processor, bool):
-            return self.configure_pre_processor() if pre_processor else None
-        msg = f"Invalid pre-processor type: {type(pre_processor)}"
-        raise TypeError(msg)
-
-    def configure_callbacks(self) -> Sequence[Callback] | Callback:
-        """Configure default callbacks for AnomalibModule."""
-        return [self.pre_processor] if self.pre_processor else []
+        callbacks: list[Callback] = []
+        callbacks.extend(
+            component
+            for component in (self.pre_processor, self.post_processor, self.evaluator, self.visualizer)
+            if isinstance(component, Callback)
+        )
+        return callbacks
 
     def forward(self, batch: torch.Tensor, *args, **kwargs) -> InferenceBatch:
         """Perform the forward-pass by passing input tensor to the module.
@@ -170,6 +165,25 @@ def learning_type(self) -> LearningType:
         """Learning type of the model."""
         raise NotImplementedError
 
+    def _resolve_pre_processor(self, pre_processor: PreProcessor | bool) -> PreProcessor | None:
+        """Resolve and validate which pre-processor to use..
+
+        Args:
+            pre_processor: Pre-processor configuration
+                - True -> use default pre-processor
+                - False -> no pre-processor
+                - PreProcessor -> use the provided pre-processor
+
+        Returns:
+            Configured pre-processor
+        """
+        if isinstance(pre_processor, PreProcessor):
+            return pre_processor
+        if isinstance(pre_processor, bool):
+            return self.configure_pre_processor() if pre_processor else None
+        msg = f"Invalid pre-processor type: {type(pre_processor)}"
+        raise TypeError(msg)
+
     @classmethod
     def configure_pre_processor(cls, image_size: tuple[int, int] | None = None) -> PreProcessor:
         """Configure the pre-processor.
@@ -214,15 +228,54 @@ def configure_pre_processor(cls, image_size: tuple[int, int] | None = None) -> P
             ]),
         )
 
-    def default_post_processor(self) -> PostProcessor | None:
-        """Default post processor.
+    def _resolve_post_processor(self, post_processor: PostProcessor | bool) -> PostProcessor | None:
+        """Resolve and validate which post-processor to use.
 
-        Override in subclass for model-specific post-processing behaviour.
+        Args:
+            post_processor: Post-processor configuration
+                - True -> use default post-processor
+                - False -> no post-processor
+                - PostProcessor -> use the provided post-processor
+
+        Returns:
+            Configured post-processor
+        """
+        if isinstance(post_processor, PostProcessor):
+            return post_processor
+        if isinstance(post_processor, bool):
+            return self.configure_post_processor() if post_processor else None
+        msg = f"Invalid post-processor type: {type(post_processor)}"
+        raise TypeError(msg)
+
+    def configure_post_processor(self) -> PostProcessor | None:
+        """Configure the default post-processor based on the learning type.
+
+        Returns:
+            PostProcessor: Configured post-processor instance.
+
+        Raises:
+            NotImplementedError: If no default post-processor is available for the model's learning type.
+
+        Examples:
+            Get default post-processor:
+
+            >>> post_processor = AnomalibModule.configure_post_processor()
+
+            Create model with custom post-processor:
+
+            >>> custom_post_processor = CustomPostProcessor()
+            >>> model = PatchCore(post_processor=custom_post_processor)
+
+            Disable post-processing:
+
+            >>> model = PatchCore(post_processor=False)
         """
         if self.learning_type == LearningType.ONE_CLASS:
             return OneClassPostProcessor()
-        msg = f"No default post-processor available for model {self.__name__} with learning type {self.learning_type}. \
-              Please override the default_post_processor method in the model implementation."
+        msg = (
+            f"No default post-processor available for model with learning type {self.learning_type}. "
+            "Please override the configure_post_processor method in the model implementation."
+        )
         raise NotImplementedError(msg)
 
     def _resolve_evaluator(self, evaluator: Evaluator | bool) -> Evaluator | None:
@@ -251,6 +304,63 @@ def configure_evaluator() -> Evaluator:
         test_metrics = [image_auroc, image_f1score, pixel_auroc, pixel_f1score]
         return Evaluator(test_metrics=test_metrics)
 
+    def _resolve_visualizer(self, visualizer: Visualizer | bool) -> Visualizer | None:
+        """Resolve and validate which visualizer to use.
+
+        Args:
+            visualizer: Visualizer configuration
+                - True -> use default visualizer
+                - False -> no visualizer
+                - Visualizer -> use the provided visualizer
+
+        Returns:
+            Configured visualizer
+        """
+        if isinstance(visualizer, Visualizer):
+            return visualizer
+        if isinstance(visualizer, bool):
+            return self.configure_visualizer() if visualizer else None
+        msg = f"Visualizer must be of type Visualizer or bool, got {type(visualizer)}"
+        raise TypeError(msg)
+
+    @classmethod
+    def configure_visualizer(cls) -> ImageVisualizer:
+        """Configure the default visualizer.
+
+        By default, this method returns an ImageVisualizer instance, which is suitable for
+        visualizing image-based anomaly detection results. However, the visualizer can be
+        customized based on your needs - for example, using VideoVisualizer for video data
+        or implementing a custom visualizer for specific visualization requirements.
+
+        Returns:
+            Visualizer: Configured visualizer instance (ImageVisualizer by default).
+
+        Examples:
+            Get default ImageVisualizer:
+
+            >>> visualizer = AnomalibModule.configure_visualizer()
+
+            Create model with VideoVisualizer:
+
+            >>> from custom_module import VideoVisualizer
+            >>> video_visualizer = VideoVisualizer()
+            >>> model = PatchCore(visualizer=video_visualizer)
+
+            Create model with custom visualizer:
+
+            >>> class CustomVisualizer(Visualizer):
+            ...     def __init__(self):
+            ...         super().__init__()
+            ...         # Custom visualization logic
+            >>> custom_visualizer = CustomVisualizer()
+            >>> model = PatchCore(visualizer=custom_visualizer)
+
+            Disable visualization:
+
+            >>> model = PatchCore(visualizer=False)
+        """
+        return ImageVisualizer()
+
     @property
     def input_size(self) -> tuple[int, int] | None:
         """Return the effective input size of the model.

src/anomalib/models/image/cfa/lightning_model.py

@@ -20,6 +20,7 @@
 from anomalib.models.components import AnomalibModule
 from anomalib.post_processing import PostProcessor
 from anomalib.pre_processing import PreProcessor
+from anomalib.visualization import Visualizer
 
 from .loss import CfaLoss
 from .torch_model import CfaModel
@@ -58,11 +59,18 @@ def __init__(
         num_nearest_neighbors: int = 3,
         num_hard_negative_features: int = 3,
         radius: float = 1e-5,
+        # Anomalib's Auxiliary Components
         pre_processor: PreProcessor | bool = True,
-        post_processor: PostProcessor | None = None,
+        post_processor: PostProcessor | bool = True,
         evaluator: Evaluator | bool = True,
+        visualizer: Visualizer | bool = True,
     ) -> None:
-        super().__init__(pre_processor=pre_processor, post_processor=post_processor, evaluator=evaluator)
+        super().__init__(
+            pre_processor=pre_processor,
+            post_processor=post_processor,
+            evaluator=evaluator,
+            visualizer=visualizer,
+        )
         self.model: CfaModel = CfaModel(
             backbone=backbone,
             gamma_c=gamma_c,

src/anomalib/models/image/cflow/lightning_model.py

@@ -27,6 +27,7 @@
 from anomalib.models.components import AnomalibModule
 from anomalib.post_processing import PostProcessor
 from anomalib.pre_processing import PreProcessor
+from anomalib.visualization import Visualizer
 
 from .torch_model import CflowModel
 from .utils import get_logp, positional_encoding_2d
@@ -71,10 +72,16 @@ def __init__(
         permute_soft: bool = False,
         lr: float = 0.0001,
         pre_processor: PreProcessor | bool = True,
-        post_processor: PostProcessor | None = None,
+        post_processor: PostProcessor | bool = True,
         evaluator: Evaluator | bool = True,
+        visualizer: Visualizer | bool = True,
     ) -> None:
-        super().__init__(pre_processor=pre_processor, post_processor=post_processor, evaluator=evaluator)
+        super().__init__(
+            pre_processor=pre_processor,
+            post_processor=post_processor,
+            evaluator=evaluator,
+            visualizer=visualizer,
+        )
 
         self.model: CflowModel = CflowModel(
             backbone=backbone,