Merge 960cce0 into d4ac3b0

deepcell/applications/application.py

@@ -138,14 +138,15 @@ def _preprocess(self, image, **kwargs):
 
         return image
 
-    def _tile_input(self, image):
+    def _tile_input(self, image, pad_mode='constant'):
         """Tile the input image to match shape expected by model
         using the ``deepcell_toolbox`` function.
 
         Only supports 4D images.
 
         Args:
             image (numpy.array): Input image to tile
+            pad_mode (str): The padding mode, one of "constant" or "reflect".
 
         Raises:
             ValueError: Input images must have only 4 dimensions
@@ -178,7 +179,8 @@ def _tile_input(self, image):
         # Otherwise tile images larger than model size
         else:
             # Tile images, needs 4d
-            tiles, tiles_info = tile_image(image, model_input_shape=self.model_image_shape)
+            tiles, tiles_info = tile_image(image, model_input_shape=self.model_image_shape,
+                                           stride_ratio=0.75, pad_mode=pad_mode)
 
         return tiles, tiles_info
 
@@ -296,12 +298,14 @@ def _resize_output(self, image, original_shape):
     def _run_model(self,
                    image,
                    batch_size=4,
+                   pad_mode='constant',
                    preprocess_kwargs={}):
         """Run the model to generate output probabilities on the data.
 
         Args:
             image (numpy.array): Image with shape ``[batch, x, y, channel]``
             batch_size (int): Number of images to predict on per batch.
+            pad_mode (str): The padding mode, one of "constant" or "reflect".
             preprocess_kwargs (dict): Keyword arguments to pass to
                 the preprocessing function.
 
@@ -313,7 +317,7 @@ def _run_model(self,
         image = self._preprocess(image, **preprocess_kwargs)
 
         # Tile images, raises error if the image is not 4d
-        tiles, tiles_info = self._tile_input(image)
+        tiles, tiles_info = self._tile_input(image, pad_mode=pad_mode)
 
         # Run images through model
         output_tiles = self.model.predict(tiles, batch_size=batch_size)
@@ -330,6 +334,7 @@ def _predict_segmentation(self,
                               image,
                               batch_size=4,
                               image_mpp=None,
+                              pad_mode='constant',
                               preprocess_kwargs={},
                               postprocess_kwargs={}):
         """Generates a labeled image of the input running prediction with
@@ -344,6 +349,7 @@ def _predict_segmentation(self,
                 ``[batch, x, y, channel]``.
             batch_size (int): Number of images to predict on per batch.
             image_mpp (float): Microns per pixel for ``image``.
+            pad_mode (str): The padding mode, one of "constant" or "reflect".
             preprocess_kwargs (dict): Keyword arguments to pass to the
                 pre-processing function.
             postprocess_kwargs (dict): Keyword arguments to pass to the
@@ -374,8 +380,10 @@ def _predict_segmentation(self,
         resized_image = self._resize_input(image, image_mpp)
 
         # Generate model outputs
-        output_images = self._run_model(image=resized_image, batch_size=batch_size,
-                                        preprocess_kwargs=preprocess_kwargs)
+        output_images = self._run_model(
+            image=resized_image, batch_size=batch_size,
+            pad_mode=pad_mode, preprocess_kwargs=preprocess_kwargs
+        )
 
         # Postprocess predictions to create label image
         label_image = self._postprocess(output_images, **postprocess_kwargs)

deepcell/applications/nuclear_segmentation.py

@@ -116,6 +116,7 @@ def predict(self,
                 image,
                 batch_size=4,
                 image_mpp=None,
+                pad_mode='reflect',
                 preprocess_kwargs=None,
                 postprocess_kwargs=None):
         """Generates a labeled image of the input running prediction with
@@ -131,6 +132,7 @@ def predict(self,
                 ``[batch, x, y, channel]``.
             batch_size (int): Number of images to predict on per batch.
             image_mpp (float): Microns per pixel for ``image``.
+            pad_mode (str): The padding mode, one of "constant" or "reflect".
             preprocess_kwargs (dict): Keyword arguments to pass to the
                 pre-processing function.
             postprocess_kwargs (dict): Keyword arguments to pass to the
@@ -164,5 +166,6 @@ def predict(self,
             image,
             batch_size=batch_size,
             image_mpp=image_mpp,
+            pad_mode=pad_mode,
             preprocess_kwargs=preprocess_kwargs,
             postprocess_kwargs=postprocess_kwargs)

requirements.txt

@@ -6,4 +6,4 @@ tensorflow==2.3.1
 jupyter>=1.0.0,<2
 opencv-python-headless<=3.4.9.31
 deepcell-tracking>=0.2.7
-deepcell-toolbox>=0.8.3
+deepcell-toolbox>=0.8.4