def test():
# Build the VGG16 network with ImageNet weights
model = VGG16(weights='imagenet', include_top=True)
# Utility to search for layer index by name.
# Alternatively we can specify this as -1 since it corresponds to the last layer.
layer_idx = utils.find_layer_idx(model, 'predictions')
# Swap softmax with linear
model.layers[layer_idx].activation = activations.linear
model = utils.apply_modifications(model)
plt.rcParams['figure.figsize'] = (18, 6)
img1 = utils.load_img('images/ouzel1.jpg', target_size=(224, 224))
img2 = utils.load_img('images/ouzel2.jpg', target_size=(224, 224))
# f, ax = plt.subplots(1, 2)
# ax[0].imshow(img1)
# ax[1].imshow(img2)
f, ax = plt.subplots(1, 2)
for i, img in enumerate([img1, img2]):
# 20 is the imagenet index corresponding to `ouzel`
# heatmap = saliency.visualize_cam(model, layer_idx, filter_indices=20, seed_input=img,backprop_modifier='guided')
heatmap = saliency.visualize_saliency(model, layer_idx, filter_indices=20, seed_input=img,backprop_modifier=None)
print (np.shape(heatmap))
# Lets overlay the heatmap onto original image.
ax[i].imshow(overlay(heatmap,img))
plt.show()
def test_for_issues_135():
inputs = Input((35, ))
x = Dense(64, activation='relu')(inputs)
x = Dense(100, activation='relu')(x)
x = Dense(50)(x)
model = Model(inputs, x)
data = np.random.rand(1, 35)
grads = visualize_saliency(model, -1, 0, data, keepdims=True)
assert grads.shape == (35, )
def test_visualize_saliency_with_unkeepdims(model, data):
grads = visualize_saliency(model, -1, 0, data, keepdims=True)
assert grads.shape == (28, 28, 3)
def test_visualize_saliency(model, data):
# FIXME Can't set None to filter_indices with Theano backend.
# To get green test, it set zero.
# grads = visualize_saliency(model, -1, filter_indices=None, seed_input=data)
grads = visualize_saliency(model, -1, filter_indices=0, seed_input=data)
assert grads.shape == (28, 28)
self = super().load(path)
self.model.layers[-1].activation = keras.activations.linear
self.model = vu.utils.apply_modifications(self.model)
return self
def get_validation_data(
self, dataset: som_dataset.SOMDataset) -> som_dataset.SOMDataset:
return dataset.filter(labels=self.data_ids["validation"])
def transform(self, case, group, maximization=False):
"""Get saliency gradients for the given group for the selected case."""
xdata, _ = self.array_from_cases([case])
input_indices = [*range(len(xdata))]
gradients = visualize_saliency(self.model,
self.layer_idx,
self.config.groups.index(group),
seed_input=xdata,
input_indices=input_indices,
maximization=maximization)
return gradients
def calculate_saliency(self,
som_sequence,
case,
group,
maximization=False):
"""Calculates the saliency values / gradients for the case, model and
each of the classes.
Args:
dataset: SOMMapDataset object.
case: Case object for which the saliency values will be computed.
group: Select group.