🛠 Fix config files and refactor dfkde

anomalib/models/dfkde/config.yaml

@@ -34,8 +34,8 @@ metrics:
 
 visualization:
   show_images: False # show images on the screen
-  save_images: True # save images to the file system
-  log_images: True # log images to the available loggers (if any)
+  save_images: False # save images to the file system
+  log_images: False # log images to the available loggers (if any)
   image_save_path: null # path to which images will be saved
   mode: full # options: ["full", "simple"]
 

anomalib/models/dfkde/torch_model.py

@@ -79,6 +79,29 @@ def get_features(self, batch: Tensor) -> Tensor:
         layer_outputs = torch.cat(list(layer_outputs.values())).detach()
         return layer_outputs
 
+    def pre_process(self, feature_stack: Tensor, max_length: Optional[Tensor] = None) -> Tuple[Tensor, Tensor]:
+        """Pre-process the CNN features.
+
+        Args:
+          feature_stack (Tensor): Features extracted from CNN
+          max_length (Optional[Tensor]): Used to unit normalize the feature_stack vector. If ``max_len`` is not
+            provided, the length is calculated from the ``feature_stack``. Defaults to None.
+
+        Returns:
+            (Tuple): Stacked features and length
+        """
+
+        if max_length is None:
+            max_length = torch.max(torch.linalg.norm(feature_stack, ord=2, dim=1))
+
+        if self.pre_processing == "norm":
+            feature_stack /= torch.linalg.norm(feature_stack, ord=2, dim=1)[:, None]
+        elif self.pre_processing == "scale":
+            feature_stack /= max_length
+        else:
+            raise RuntimeError("Unknown pre-processing mode. Available modes are: Normalized and Scale.")
+        return feature_stack, max_length
+
     def fit(self, embeddings: List[Tensor]) -> bool:
         """Fit a kde model to embeddings.
 
@@ -103,36 +126,13 @@ def fit(self, embeddings: List[Tensor]) -> bool:
             selected_features = _embeddings
 
         feature_stack = self.pca_model.fit_transform(selected_features)
-        feature_stack, max_length = self.preprocess(feature_stack)
+        feature_stack, max_length = self.pre_process(feature_stack)
         self.max_length = max_length
         self.kde_model.fit(feature_stack)
 
         return True
 
-    def preprocess(self, feature_stack: Tensor, max_length: Optional[Tensor] = None) -> Tuple[Tensor, Tensor]:
-        """Pre-process the CNN features.
-
-        Args:
-          feature_stack (Tensor): Features extracted from CNN
-          max_length (Optional[Tensor]): Used to unit normalize the feature_stack vector. If ``max_len`` is not
-            provided, the length is calculated from the ``feature_stack``. Defaults to None.
-
-        Returns:
-            (Tuple): Stacked features and length
-        """
-
-        if max_length is None:
-            max_length = torch.max(torch.linalg.norm(feature_stack, ord=2, dim=1))
-
-        if self.pre_processing == "norm":
-            feature_stack /= torch.linalg.norm(feature_stack, ord=2, dim=1)[:, None]
-        elif self.pre_processing == "scale":
-            feature_stack /= max_length
-        else:
-            raise RuntimeError("Unknown pre-processing mode. Available modes are: Normalized and Scale.")
-        return feature_stack, max_length
-
-    def evaluate(self, features: Tensor, as_log_likelihood: Optional[bool] = False) -> Tensor:
+    def compute_kde_scores(self, features: Tensor, as_log_likelihood: Optional[bool] = False) -> Tensor:
         """Compute the KDE scores.
 
         The scores calculated from the KDE model are converted to densities. If `as_log_likelihood` is set to true then
@@ -147,7 +147,7 @@ def evaluate(self, features: Tensor, as_log_likelihood: Optional[bool] = False)
         """
 
         features = self.pca_model.transform(features)
-        features, _ = self.preprocess(features, self.max_length)
+        features, _ = self.pre_process(features, self.max_length)
         # Scores are always assumed to be passed as a density
         kde_scores = self.kde_model(features)
 
@@ -159,32 +159,32 @@ def evaluate(self, features: Tensor, as_log_likelihood: Optional[bool] = False)
 
         return kde_scores
 
-    def predict(self, features: Tensor) -> Tensor:
-        """Predicts the probability that the features belong to the anomalous class.
+    def compute_probabilities(self, scores: Tensor) -> Tensor:
+        """Converts density scores to anomaly probabilities (see https://www.desmos.com/calculator/ifju7eesg7).
 
         Args:
-          features (Tensor): Feature from which the output probabilities are detected.
+          scores (Tensor): density of an image.
 
         Returns:
-          Detection probabilities
+          probability that image with {density} is anomalous
         """
 
-        densities = self.evaluate(features, as_log_likelihood=True)
-        probabilities = self.to_probability(densities)
-
-        return probabilities
+        return 1 / (1 + torch.exp(self.threshold_steepness * (scores - self.threshold_offset)))
 
-    def to_probability(self, densities: Tensor) -> Tensor:
-        """Converts density scores to anomaly probabilities (see https://www.desmos.com/calculator/ifju7eesg7).
+    def predict(self, features: Tensor) -> Tensor:
+        """Predicts the probability that the features belong to the anomalous class.
 
         Args:
-          densities (Tensor): density of an image.
+          features (Tensor): Feature from which the output probabilities are detected.
 
         Returns:
-          probability that image with {density} is anomalous
+          Detection probabilities
         """
 
-        return 1 / (1 + torch.exp(self.threshold_steepness * (densities - self.threshold_offset)))
+        scores = self.compute_kde_scores(features, as_log_likelihood=True)
+        probabilities = self.compute_probabilities(scores)
+
+        return probabilities
 
     def forward(self, batch: Tensor) -> Tensor:
         """Prediction by normality model.

anomalib/models/dfm/config.yaml

@@ -36,8 +36,8 @@ metrics:
 
 visualization:
   show_images: False # show images on the screen
-  save_images: True # save images to the file system
-  log_images: True # log images to the available loggers (if any)
+  save_images: False # save images to the file system
+  log_images: False # log images to the available loggers (if any)
   image_save_path: null # path to which images will be saved
   mode: full # options: ["full", "simple"]