patch duplicate function (#73)

Co-authored-by: Sam <sam.vanhoutte@marchitec.be>

Author: SamVanhoutte

arcus/azureml/experimenting/aml_trainer.py

@@ -286,25 +286,6 @@ def evaluate_image_classifier(self, fitted_model, X_test: np.array, y_test: np.a
         if return_predictions:  
             return y_pred
 
-    def save_image_outputs(self, X_test: np.array, y_test: np.array, y_pred: np.array, samples_to_save: int = 1) -> np.array:
-        '''
-        Will save image outputs to the run
-        Args:
-            X_test (np.array): The input images for the model
-            y_test (np.array): The actual expected output images of the model
-            y_pred (np.array): The predicted or calculated output images of the model
-            samples_to_save (int): If greather than 0, this amount of input, output and generated image combinations will be tracked to the Run
-        ''' 
-
-        if samples_to_save > 0:
-            import random
-            total_images = min(len(y_pred), samples_to_save)
-
-            for i in random.sample(range(len(y_pred)), total_images):
-                newimg = self.concat_images([X_test[i], y_test[i], y_pred[i]])
-                imgplot = explorer.show_image(newimg, silent_mode=True)
-                self.__current_run.log_image(f'Image combo sample {i}', plot=imgplot)
-                imgplot.close()
 
 
     def __stack_images(self, img1: np.array, img2: np.array):
@@ -365,7 +346,6 @@ def evaluate_image_classifier(self, fitted_model, X_test: np.array, y_test: np.a
             return y_pred
 
     def save_image_outputs(self, X_test: np.array, y_test: np.array, y_pred: np.array, samples_to_save: int = 1) -> np.array:
-
         '''
         Will save image outputs to the run
         Args:
@@ -376,15 +356,36 @@ def save_image_outputs(self, X_test: np.array, y_test: np.array, y_pred: np.arra
         ''' 
 
         if samples_to_save > 0:
-            # Take incorrect classified images and save
             import random
             total_images = min(len(y_pred), samples_to_save)
 
             for i in random.sample(range(len(y_pred)), total_images):
-                groupplot = explorer.visualize({'Charts': [X_test[i]], 'Actuals': [y_test[i]], 'Calculated': [y_pred[i]]}, 1, grid_size=(6, 6), silent_mode=True)
-                image = X_test[i].reshape(X_test.shape[1], X_test.shape[2])
-                imgplot = explorer.show_image(image, silent_mode=True)
-                self.__current_run.log_image(f'Sample {i:02d} / {total_images:02d}', plot=groupplot)
+                newimg = self.concat_images([X_test[i], y_test[i], y_pred[i]])
+                imgplot = explorer.show_image(newimg, silent_mode=True)
+                self.__current_run.log_image(f'Image combo sample {i}', plot=imgplot)
+                imgplot.close()
+
+    # def save_image_outputs(self, X_test: np.array, y_test: np.array, y_pred: np.array, samples_to_save: int = 1) -> np.array:
+
+    #     '''
+    #     Will save image outputs to the run
+    #     Args:
+    #         X_test (np.array): The input images for the model
+    #         y_test (np.array): The actual expected output images of the model
+    #         y_pred (np.array): The predicted or calculated output images of the model
+    #         samples_to_save (int): If greather than 0, this amount of input, output and generated image combinations will be tracked to the Run
+    #     ''' 
+
+    #     if samples_to_save > 0:
+    #         # Take incorrect classified images and save
+    #         import random
+    #         total_images = min(len(y_pred), samples_to_save)
+
+    #         for i in random.sample(range(len(y_pred)), total_images):
+    #             groupplot = explorer.visualize({'Charts': [X_test[i]], 'Actuals': [y_test[i]], 'Calculated': [y_pred[i]]}, 1, grid_size=(6, 6), silent_mode=True)
+    #             image = X_test[i].reshape(X_test.shape[1], X_test.shape[2])
+    #             imgplot = explorer.show_image(image, silent_mode=True)
+    #             self.__current_run.log_image(f'Sample {i:02d} / {total_images:02d}', plot=groupplot)
 
     def setup_training(self, training_name: str, overwrite: bool = False):
         '''