-
Notifications
You must be signed in to change notification settings - Fork 0
/
grad-cam.py
194 lines (159 loc) · 6.5 KB
/
grad-cam.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
from keras.applications.vgg16 import VGG16, preprocess_input, decode_predictions
from keras.preprocessing import image
from keras.layers.core import Lambda
from keras.models import Sequential, load_model
from tensorflow.python.framework import ops
import keras.backend as K
import tensorflow as tf
import numpy as np
import keras
import sys
import cv2
import imutils
import h5py
import pickle
def target_category_loss(x, category_index, nb_classes):
return tf.multiply(x, K.one_hot([category_index], nb_classes))
def target_category_loss_output_shape(input_shape):
return input_shape
def normalize(x):
# utility function to normalize a tensor by its L2 norm
return x / (K.sqrt(K.mean(K.square(x))) + 1e-5)
def register_gradient():
if "GuidedBackProp" not in ops._gradient_registry._registry:
@ops.RegisterGradient("GuidedBackProp")
def _GuidedBackProp(op, grad):
dtype = op.inputs[0].dtype
return grad * tf.cast(grad > 0., dtype) * \
tf.cast(op.inputs[0] > 0., dtype)
def compile_saliency_function(model, activation_layer='conv2d_3'):
input_img = model.input
#layer_dict = dict([(layer.name, layer) for layer in model.layers[1:]])
layer_dict = dict([(layer.name, layer) for layer in model.layers])
#print("layer_dict", layer_dict)
layer_output = layer_dict[activation_layer].output
#print("layer_output", layer_output)
max_output = K.max(layer_output, axis=3)
#print("max_output", max_output)
saliency = K.gradients(K.sum(max_output), input_img)[0]
return K.function([input_img, K.learning_phase()], [saliency])
def modify_backprop(model, name):
g = tf.get_default_graph()
with g.gradient_override_map({'Relu': name}):
# get layers that have an activation
layer_dict = [layer for layer in model.layers[1:]
if hasattr(layer, 'activation')]
# replace relu activation
for layer in layer_dict:
if layer.activation == keras.activations.relu:
layer.activation = tf.nn.relu
# re-instanciate a new model
#new_model = VGG16(weights='imagenet')
new_model = load_model('.../gradCam-smallvgg/pokedex.model')
return new_model
def deprocess_image(x):
'''
Same normalization as in:
https://github.com/fchollet/keras/blob/master/examples/conv_filter_visualization.py
'''
if np.ndim(x) > 3:
x = np.squeeze(x)
# normalize tensor: center on 0., ensure std is 0.1
x -= x.mean()
x /= (x.std() + 1e-5)
x *= 0.1
# clip to [0, 1]
x += 0.5
x = np.clip(x, 0, 1)
# convert to RGB array
x *= 255
if K.image_dim_ordering() == 'th':
x = x.transpose((1, 2, 0))
x = np.clip(x, 0, 255).astype('uint8')
return x
def grad_cam(input_model, image, category_index, layer_name):
nb_classes = 5
#nb_classes = 1000
target_layer = lambda x: target_category_loss(x, category_index, nb_classes)
x = input_model.output
#x = input_model.layers[-1].output
x = Lambda(target_layer, output_shape=target_category_loss_output_shape)(x)
#model = keras.models.Model(input_model.layers[0].input, x)
model = keras.models.Model(inputs=input_model.input, outputs=x)
#loss = K.sum(model.layers[-1].output)
loss = K.sum(model.output)
#compute_heatmap
# For VGG16
#conv_output = [l for l in model.layers[0].layers if l.name is layer_name][0].output
#conv_output = [l for l in model.layers if l.name is layer_name][0].output
conv_output = model.get_layer(layer_name).output
#print("conv_output", conv_output)
grads = normalize(K.gradients(loss, conv_output)[0])
gradient_function = K.function([model.layers[0].input], [conv_output, grads])
output, grads_val = gradient_function([image])
output, grads_val = output[0, :], grads_val[0, :, :, :]
weights = np.mean(grads_val, axis = (0, 1))
cam = np.ones(output.shape[0 : 2], dtype = np.float32)
for i, w in enumerate(weights):
cam += w * output[:, :, i]
#cam = cv2.resize(cam, (224, 224)) # For VGG16
cam = cv2.resize(cam, (96, 96))
cam = np.maximum(cam, 0)
heatmap = cam / np.max(cam)
#Return to BGR [0..255] from the preprocessed image
image = image[0, :]
image -= np.min(image)
image = np.minimum(image, 255)
cam = cv2.applyColorMap(np.uint8(255*heatmap), cv2.COLORMAP_JET)
cam = np.float32(cam) + np.float32(image)
cam = 255 * cam / np.max(cam)
heatmap = cv2.applyColorMap(np.uint8(255*heatmap), cv2.COLORMAP_JET)
return np.uint8(cam), heatmap
img_path = sys.argv[1]
orig = cv2.imread(img_path)
img = cv2.resize(orig, (96, 96))
#img = cv2.resize(orig, (224, 224)) # For VGG16
img = img.astype("float") / 255.0
#img = image.load_img(img_path, target_size=(224, 224)) # For VGG16
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
#x = preprocess_input(x)
#model = VGG16(weights='imagenet') # For VGG16
print("[INFO] loading network...")
model = load_model('.../gradCam-smallvgg/pokedex.model')
lb = pickle.loads(open(".../gradCam-smallvgg/lb.pickle", "rb").read())
model.summary()
# classify the input image
print("[INFO] classifying image...")
predictions = model.predict(x)#[0]
idx = np.argmax(predictions)
label = lb.classes_[idx]
label = "{}: {:.2f}%".format(label, predictions[0][idx] * 100)
#print("idx", idx)
print("label", label)
# For VGG16
#top_1 = decode_predictions(predictions)
#(imagenetID, label, prob) = top_1[0][0]
#label = "{}: {:.2f}%".format(label, prob * 100)
#print("[INFO] {}".format(label))
#predicted_class = np.argmax(predictions)
our, heatmap = grad_cam(model, x, idx, "dropout_3")
#our, heatmap = grad_cam(model, x, predicted_class, "block5_conv3") # For VGG16
heatmap2 = cv2.resize(heatmap, (orig.shape[1], orig.shape[0]))
our = cv2.resize(our, (orig.shape[1], orig.shape[0]))
cv2.rectangle(our, (0, 0), (340, 40), (0, 0, 0), -1)
cv2.putText(our, label, (10, 25), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 255, 255), 2)
register_gradient()
guided_model = modify_backprop(model, 'GuidedBackProp')
saliency_fn = compile_saliency_function(guided_model, "dropout_3")
saliency = saliency_fn([x, 0])
#gradcam = saliency[0] * heatmap[..., np.newaxis] # For VGG16
gradcam = saliency[0] * heatmap[np.newaxis, ...]
gradcam = deprocess_image(gradcam)
gradcam = cv2.resize(gradcam, (orig.shape[1], orig.shape[0]))
#cv2.imwrite("guided_gradcam1.jpg", deprocess_image(gradcam))
our = np.vstack([orig, our, gradcam])
our = imutils.resize(our, height=900)
cv2.imshow("gradcam", our)
cv2.waitKey(0)
cv2.imwrite("dropout_3.jpg", our)