Merge pull request #73 from keisen/releases/v0.8.0

Releases/v0.8.0

LICENSE

@@ -1,7 +1,7 @@
 MIT License
 
 Copyright (c) 2016 Raghavendra Kotikalapudi
-Copyright (c) 2019 keisen(Kubota Yasuhiro) and contributors
+Copyright (c) 2019 keisen(Yasuhiro Kubota) and contributors
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

MANIFEST.in

@@ -0,0 +1,2 @@
+include VERSION
+recursive-exclude tests *

VERSION

@@ -1 +1 @@
-0.7.2
+0.8.0

tests/tf_keras_vis/activation_maximization/activation_maximization_test.py

@@ -199,12 +199,6 @@ def test__call__if_regularizers_are_(self, regularizer_container, regularizers,
             result = activation_maximization(CategoricalScore(0), regularizers=regularizers)
             assert result.shape == (1, 8, 8, 3)
 
-    @pytest.mark.usefixtures("mixed_precision", "legacy")
-    def test__call__if_normalize_gradient_is_True(self, conv_model):
-        activation_maximization = ActivationMaximization(conv_model)
-        result = activation_maximization(CategoricalScore(0), normalize_gradient=True)
-        assert result.shape == (1, 8, 8, 3)
-
     @pytest.mark.usefixtures("mixed_precision", "legacy")
     def test__call__with_gradient_modifier(self, conv_model):
         activation_maximization = ActivationMaximization(conv_model)

tests/tf_keras_vis/gradcam_test.py

@@ -1,13 +1,2 @@
-import pytest
-
-from tf_keras_vis.gradcam import Gradcam
-from tf_keras_vis.utils.scores import CategoricalScore
-from tf_keras_vis.utils.test import dummy_sample
-
-
 class TestGradcam():
-    @pytest.mark.usefixtures("mixed_precision")
-    def test__call__if_normalize_gradient_is_True(self, conv_model):
-        cam = Gradcam(conv_model)
-        result = cam(CategoricalScore(0), dummy_sample((1, 8, 8, 3)), normalize_gradient=True)
-        assert result.shape == (1, 8, 8)
+    pass

tf_keras_vis/__init__.py

@@ -1,3 +1,10 @@
+try:
+    from importlib.metadata import version
+except ImportError:  # pragma: no cover
+    from importlib_metadata import version
+
+__version__ = version("tf-keras-vis")
+
 from abc import ABC, abstractmethod
 from typing import Union
 
@@ -6,21 +13,14 @@
 
 from .utils import listify
 
-try:
-    from importlib.metadata import version
-except ImportError:  # pragma: no cover
-    from importlib_metadata import version
-
-__version__ = version("tf-keras-vis")
-
 
 class ModelVisualization(ABC):
     """Visualization class that analyze the model for debugging.
     """
     def __init__(self, model, model_modifier=None, clone=True) -> None:
         """
         Args:
-            model: A  `tf.keras.Model` instance. When `model_modifier` is NOT None, this model will
+            model: A `tf.keras.Model` instance. When `model_modifier` is NOT None, this model will
                 be cloned with `tf.keras.models.clone_model` function and then will be modified by
                 `model_modifier` according to needs.
             model_modifier: A :obj:`tf_keras_vis.utils.model_modifiers.ModelModifier` instance,

tf_keras_vis/activation_maximization/__init__.py

@@ -27,7 +27,6 @@ def __call__(
                           Norm(weight=0.3, p=1)],
             steps=200,
             optimizer=None,  # When None, the default is tf.optimizers.RMSprop(1.0, 0.999)
-            normalize_gradient=None,  # Disabled option.
             gradient_modifier=None,
             callbacks=None,
             training=False,
@@ -150,8 +149,7 @@ def __call__(
             gradient_modifier: A function to modify gradients.
                 Defaults to None.
             callbacks: A :obj:`tf_keras_vis.activation_maximization.callbacks.Callback` instance
-                or a list of them.
-                Defaults to None.
+                or a list of them. Defaults to None.
             training: A bool that indicates whether the model's training-mode on or off.
                 Defaults to False.
             unconnected_gradients: Specifies the gradient value returned when the given input
@@ -177,11 +175,6 @@ def __call__(
         arguments = dict(
             (k, v) for k, v in locals().items() if k != 'self' and not k.startswith('_'))
 
-        if normalize_gradient is not None:
-            warnings.warn(
-                "`normalize_gradient` option of ActivationMaximization#__call__() is disabled.,"
-                " And this will be removed in future.", DeprecationWarning)
-
         # Check model
         mixed_precision_model = is_mixed_precision(self.model)
 

tf_keras_vis/activation_maximization/legacy.py

@@ -18,7 +18,6 @@ def __call__(self,
                                Norm(weight=0.003, p=1)],
                  steps=200,
                  optimizer=None,
-                 normalize_gradient=None,
                  gradient_modifier=None,
                  callbacks=None,
                  training=False,

tf_keras_vis/gradcam.py

@@ -1,4 +1,3 @@
-import warnings
 from typing import Union
 
 import numpy as np
@@ -7,7 +6,7 @@
 from scipy.ndimage.interpolation import zoom
 
 from . import ModelVisualization
-from .utils import is_mixed_precision, standardize, zoom_factor
+from .utils import is_mixed_precision, normalize, zoom_factor
 from .utils.model_modifiers import ExtractIntermediateLayerForGradcam as ModelModifier
 
 
@@ -19,19 +18,17 @@ class Gradcam(ModelVisualization):
           Visual Explanations from Deep Networks via Gradient-based Localization
           (https://arxiv.org/pdf/1610.02391v1.pdf)
     """
-    def __call__(
-            self,
-            score,
-            seed_input,
-            penultimate_layer=None,
-            seek_penultimate_conv_layer=True,
-            gradient_modifier=None,
-            activation_modifier=lambda cam: K.relu(cam),
-            training=False,
-            normalize_gradient=None,  # Disabled option.
-            expand_cam=True,
-            standardize_cam=True,
-            unconnected_gradients=tf.UnconnectedGradients.NONE) -> Union[np.ndarray, list]:
+    def __call__(self,
+                 score,
+                 seed_input,
+                 penultimate_layer=None,
+                 seek_penultimate_conv_layer=True,
+                 gradient_modifier=None,
+                 activation_modifier=lambda cam: K.relu(cam),
+                 training=False,
+                 expand_cam=True,
+                 normalize_cam=True,
+                 unconnected_gradients=tf.UnconnectedGradients.NONE) -> Union[np.ndarray, list]:
         """Generate gradient based class activation maps (CAM) by using positive gradient of
         penultimate_layer with respect to score.
 
@@ -64,16 +61,14 @@ def __call__(
             seek_penultimate_conv_layer: A bool that indicates whether or not seeks a penultimate
                 layer when the layer specified by `penultimate_layer` is not `convolutional` layer.
                 Defaults to True.
+            gradient_modifier: A function to modify gradients. Defaults to None.
             activation_modifier: A function to modify the Class Activation Map (CAM). Defaults to
                 `lambda cam: K.relu(cam)`.
             training: A bool that indicates whether the model's training-mode on or off. Defaults
                 to False.
-            normalize_gradient (bool, optional): **Note!** This option is now disabled.
-                Defaults to None.
-            gradient_modifier: A function to modify gradients. Defaults to None.
             expand_cam: True to resize CAM to the same as input image size. **Note!** When False,
                 even if the model has multiple inputs, return only a CAM. Defaults to True.
-            standardize_cam: When True, CAM will be standardized. Defaults to True.
+            normalize_cam: When True, CAM will be normalized. Defaults to True.
             unconnected_gradients: Specifies the gradient value returned when the given input
                 tensors are unconnected. Defaults to tf.UnconnectedGradients.NONE.
 
@@ -86,10 +81,6 @@ def __call__(
             :obj:`ValueError`: When there is any invalid arguments.
         """
 
-        if normalize_gradient is not None:
-            warnings.warn(
-                '`normalize_gradient` option is disabled.,'
-                ' And this will be removed in future.', DeprecationWarning)
         # Preparing
         scores = self._get_scores_for_multiple_outputs(score)
         seed_inputs = self._get_seed_inputs_for_multiple_inputs(seed_input)
@@ -119,15 +110,15 @@ def __call__(
             cam = activation_modifier(cam)
 
         if not expand_cam:
-            if standardize_cam:
-                cam = standardize(cam)
+            if normalize_cam:
+                cam = normalize(cam)
             return cam
 
         # Visualizing
         factors = (zoom_factor(cam.shape, X.shape) for X in seed_inputs)
         cam = [zoom(cam, factor, order=1) for factor in factors]
-        if standardize_cam:
-            cam = [standardize(x) for x in cam]
+        if normalize_cam:
+            cam = [normalize(x) for x in cam]
         if len(self.model.inputs) == 1 and not isinstance(seed_input, list):
             cam = cam[0]
         return cam

tf_keras_vis/gradcam_plus_plus.py

@@ -6,7 +6,7 @@
 from scipy.ndimage.interpolation import zoom
 
 from . import ModelVisualization
-from .utils import is_mixed_precision, standardize, zoom_factor
+from .utils import is_mixed_precision, normalize, zoom_factor
 from .utils.model_modifiers import ExtractIntermediateLayerForGradcam as ModelModifier
 
 
@@ -26,7 +26,7 @@ def __call__(self,
                  activation_modifier=lambda cam: K.relu(cam),
                  training=False,
                  expand_cam=True,
-                 standardize_cam=True,
+                 normalize_cam=True,
                  unconnected_gradients=tf.UnconnectedGradients.NONE) -> Union[np.ndarray, list]:
         """Generate gradient based class activation maps (CAM) by using positive gradient of
             penultimate_layer with respect to score.
@@ -67,7 +67,7 @@ def __call__(self,
                 to False.
             expand_cam: True to resize CAM to the same as input image size. **Note!** When False,
                 even if the model has multiple inputs, return only a CAM. Defaults to True.
-            standardize_cam: When True, CAM will be standardized. Defaults to True.
+            normalize_cam: When True, CAM will be normalized. Defaults to True.
             unconnected_gradients: Specifies the gradient value returned when the given input
                 tensors are unconnected. Defaults to tf.UnconnectedGradients.NONE.
 
@@ -141,15 +141,15 @@ def __call__(self,
             cam = activation_modifier(cam)
 
         if not expand_cam:
-            if standardize_cam:
-                cam = standardize(cam)
+            if normalize_cam:
+                cam = normalize(cam)
             return cam
 
         # Visualizing
         factors = (zoom_factor(cam.shape, X.shape) for X in seed_inputs)
         cam = [zoom(cam, factor, order=1) for factor in factors]
-        if standardize_cam:
-            cam = [standardize(x) for x in cam]
+        if normalize_cam:
+            cam = [normalize(x) for x in cam]
         if len(self.model.inputs) == 1 and not isinstance(seed_input, list):
             cam = cam[0]
         return cam

tf_keras_vis/layercam.py

@@ -23,7 +23,7 @@ def __call__(self,
                  activation_modifier=lambda cam: K.relu(cam),
                  training=False,
                  expand_cam=True,
-                 standardize_cam=True,
+                 normalize_cam=True,
                  unconnected_gradients=tf.UnconnectedGradients.NONE) -> Union[np.ndarray, list]:
         """Generate gradient based class activation maps (CAM) by using positive gradient of
         penultimate_layer with respect to score.
@@ -65,7 +65,7 @@ def __call__(self,
                 `lambda grads: tf.keras.backend.relu(grads)`.
             expand_cam: True to resize CAM to the same as input image size. **Note!** When False,
                 even if the model has multiple inputs, return only a CAM. Defaults to True.
-            standardize_cam: When True, CAM will be standardized. Defaults to True.
+            normalize_cam: When True, CAM will be normalized. Defaults to True.
             unconnected_gradients: Specifies the gradient value returned when the given input
                 tensors are unconnected. Defaults to tf.UnconnectedGradients.NONE.
 

tf_keras_vis/saliency.py

@@ -5,7 +5,7 @@
 import tensorflow.keras.backend as K
 
 from . import ModelVisualization
-from .utils import get_num_of_steps_allowed, listify, standardize
+from .utils import get_num_of_steps_allowed, listify, normalize
 
 
 class Saliency(ModelVisualization):
@@ -24,7 +24,7 @@ def __call__(self,
                  keepdims=False,
                  gradient_modifier=lambda grads: K.abs(grads),
                  training=False,
-                 standardize_saliency=True,
+                 normalize_map=True,
                  unconnected_gradients=tf.UnconnectedGradients.NONE) -> Union[np.ndarray, list]:
         """Generate an attention map that appears how output value changes with respect to a small
         change in input image pixels.
@@ -60,7 +60,7 @@ def __call__(self,
             gradient_modifier: A function to modify gradients. Defaults to None.
             training: A bool that indicates whether the model's training-mode on or off. Defaults
                 to False.
-            standardize_saliency (bool, optional): When True, saliency map will be standardized.
+            normalize_map (bool, optional): When True, saliency map will be normalized.
                 Defaults to True.
             unconnected_gradients: Specifies the gradient value returned when the given input
                 tensors are unconnected. Defaults to tf.UnconnectedGradients.NONE.
@@ -102,8 +102,8 @@ def __call__(self,
         # Visualizing
         if not keepdims:
             grads = [np.max(g, axis=-1) for g in grads]
-        if standardize_saliency:
-            grads = [standardize(g) for g in grads]
+        if normalize_map:
+            grads = [normalize(g) for g in grads]
         if len(self.model.inputs) == 1 and not isinstance(seed_input, list):
             grads = grads[0]
         return grads

tf_keras_vis/scorecam.py

@@ -6,7 +6,7 @@
 from scipy.ndimage.interpolation import zoom
 
 from . import ModelVisualization
-from .utils import get_num_of_steps_allowed, is_mixed_precision, listify, standardize, zoom_factor
+from .utils import get_num_of_steps_allowed, is_mixed_precision, listify, normalize, zoom_factor
 from .utils.model_modifiers import ExtractIntermediateLayerForGradcam as ModelModifier
 
 
@@ -28,7 +28,7 @@ def __call__(self,
                  max_N=None,
                  training=False,
                  expand_cam=True,
-                 standardize_cam=True) -> Union[np.ndarray, list]:
+                 normalize_cam=True) -> Union[np.ndarray, list]:
         """Generate score-weighted class activation maps (CAM) by using gradient-free
         visualization method.
 
@@ -71,7 +71,7 @@ def __call__(self,
                 to False.
             expand_cam: True to resize CAM to the same as input image size. **Note!** When False,
                 even if the model has multiple inputs, return only a CAM. Defaults to True.
-            standardize_cam: When True, CAM will be standardized. Defaults to True.
+            normalize_cam: When True, CAM will be normalized. Defaults to True.
             unconnected_gradients: Specifies the gradient value returned when the given input
                 tensors are unconnected. Defaults to tf.UnconnectedGradients.NONE.
 
@@ -177,15 +177,15 @@ def __call__(self,
             cam = activation_modifier(cam)
 
         if not expand_cam:
-            if standardize_cam:
-                cam = standardize(cam)
+            if normalize_cam:
+                cam = normalize(cam)
             return cam
 
         # Visualizing
         zoom_factors = (zoom_factor(cam.shape, X.shape) for X in seed_inputs)
         cam = [zoom(cam, factor, order=1) for factor in zoom_factors]
-        if standardize_cam:
-            cam = [standardize(x) for x in cam]
+        if normalize_cam:
+            cam = [normalize(x) for x in cam]
         if len(self.model.inputs) == 1 and not isinstance(seed_input, list):
             cam = cam[0]
         return cam