visualkeras: custom vgg16 show dimension example#
An example showing the visualkeras function
used by a tf.keras.Model model.
9 # Authors: The scikit-plots developers
10 # SPDX-License-Identifier: BSD-3-Clause
pip install protobuf==5.29.4
14 import tensorflow as tf
15
16 # Clear any session to reset the state of TensorFlow/Keras
17 tf.keras.backend.clear_session()
18
19 from scikitplot import visualkeras
create VGG16
23 image_size = 224
24 model = tf.keras.models.Sequential()
25 model.add(tf.keras.layers.InputLayer(shape=(image_size, image_size, 3)))
26
27 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
28 model.add(tf.keras.layers.Conv2D(64, activation="relu", kernel_size=(3, 3)))
29 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
30 model.add(tf.keras.layers.Conv2D(64, activation="relu", kernel_size=(3, 3)))
31 model.add(visualkeras.SpacingDummyLayer())
32
33 model.add(tf.keras.layers.MaxPooling2D((2, 2), strides=(2, 2)))
34 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
35 model.add(tf.keras.layers.Conv2D(128, activation="relu", kernel_size=(3, 3)))
36 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
37 model.add(tf.keras.layers.Conv2D(128, activation="relu", kernel_size=(3, 3)))
38 model.add(visualkeras.SpacingDummyLayer())
39
40 model.add(tf.keras.layers.MaxPooling2D((2, 2), strides=(2, 2)))
41 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
42 model.add(tf.keras.layers.Conv2D(256, activation="relu", kernel_size=(3, 3)))
43 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
44 model.add(tf.keras.layers.Conv2D(256, activation="relu", kernel_size=(3, 3)))
45 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
46 model.add(tf.keras.layers.Conv2D(256, activation="relu", kernel_size=(3, 3)))
47 model.add(visualkeras.SpacingDummyLayer())
48
49 model.add(tf.keras.layers.MaxPooling2D((2, 2), strides=(2, 2)))
50 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
51 model.add(tf.keras.layers.Conv2D(512, activation="relu", kernel_size=(3, 3)))
52 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
53 model.add(tf.keras.layers.Conv2D(512, activation="relu", kernel_size=(3, 3)))
54 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
55 model.add(tf.keras.layers.Conv2D(512, activation="relu", kernel_size=(3, 3)))
56 model.add(visualkeras.SpacingDummyLayer())
57
58 model.add(tf.keras.layers.MaxPooling2D((2, 2), strides=(2, 2)))
59 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
60 model.add(tf.keras.layers.Conv2D(512, activation="relu", kernel_size=(3, 3)))
61 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
62 model.add(tf.keras.layers.Conv2D(512, activation="relu", kernel_size=(3, 3)))
63 model.add(tf.keras.layers.ZeroPadding2D((1, 1)))
64 model.add(tf.keras.layers.Conv2D(512, activation="relu", kernel_size=(3, 3)))
65 model.add(tf.keras.layers.MaxPooling2D())
66 model.add(visualkeras.SpacingDummyLayer())
67
68 model.add(tf.keras.layers.Flatten())
69
70 model.add(tf.keras.layers.Dense(4096, activation="relu"))
71 model.add(tf.keras.layers.Dropout(0.5))
72 model.add(tf.keras.layers.Dense(4096, activation="relu"))
73 model.add(tf.keras.layers.Dropout(0.5))
74 model.add(tf.keras.layers.Dense(1000, activation="softmax"))
75 # model.summary()
Now visualize the model!
80 from collections import defaultdict
81
82 color_map = defaultdict(dict)
83 color_map[tf.keras.layers.Conv2D]["fill"] = "orange"
84 color_map[tf.keras.layers.ZeroPadding2D]["fill"] = "gray"
85 color_map[tf.keras.layers.Dropout]["fill"] = "pink"
86 color_map[tf.keras.layers.MaxPooling2D]["fill"] = "red"
87 color_map[tf.keras.layers.Dense]["fill"] = "green"
88 color_map[tf.keras.layers.Flatten]["fill"] = "teal"
91 from PIL import ImageFont
92
93 ImageFont.load_default()
<PIL.ImageFont.FreeTypeFont object at 0x7f84141c92d0>
97 img_vgg16_show_dimension = visualkeras.layered_view(
98 model,
99 legend=True,
100 show_dimension=True,
101 type_ignore=[visualkeras.SpacingDummyLayer],
102 font={
103 "font_size": 61,
104 # 'use_default_font': False,
105 # 'font_path': '/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf'
106 },
107 # to_file="result_images/vgg16_show_dimension.png",
108 save_fig=True,
109 save_fig_filename="vgg16_show_dimension.png",
110 )
111 img_vgg16_show_dimension
<matplotlib.image.AxesImage object at 0x7f841400d510>
114 img_vgg16_legend_show_dimension = visualkeras.layered_view(
115 model,
116 legend=True,
117 show_dimension=True,
118 type_ignore=[visualkeras.SpacingDummyLayer],
119 font={
120 "font_size": 61,
121 # 'use_default_font': False,
122 # 'font_path': '/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf'
123 },
124 # to_file="result_images/vgg16_legend_show_dimension.png",
125 save_fig=True,
126 save_fig_filename="vgg16_legend_show_dimension.png",
127 )
128 img_vgg16_legend_show_dimension
<matplotlib.image.AxesImage object at 0x7f84140c9650>
131 img_vgg16_spacing_layers_show_dimension = visualkeras.layered_view(
132 model,
133 legend=True,
134 show_dimension=True,
135 type_ignore=[],
136 spacing=0,
137 font={
138 "font_size": 61,
139 # 'use_default_font': False,
140 # 'font_path': '/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf'
141 },
142 # to_file="result_images/vgg16_spacing_layers_show_dimension.png",
143 save_fig=True,
144 save_fig_filename="vgg16_spacing_layers_show_dimension.png",
145 )
146 img_vgg16_spacing_layers_show_dimension
<matplotlib.image.AxesImage object at 0x7f84140c2290>
149 img_vgg16_type_ignore_show_dimension = visualkeras.layered_view(
150 model,
151 legend=True,
152 show_dimension=True,
153 type_ignore=[
154 tf.keras.layers.ZeroPadding2D,
155 tf.keras.layers.Dropout,
156 tf.keras.layers.Flatten,
157 visualkeras.SpacingDummyLayer,
158 ],
159 font={
160 "font_size": 61,
161 # 'use_default_font': False,
162 # 'font_path': '/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf'
163 },
164 # to_file="result_images/vgg16_type_ignore_show_dimension.png",
165 save_fig=True,
166 save_fig_filename="vgg16_type_ignore_show_dimension.png",
167 )
168 img_vgg16_type_ignore_show_dimension
<matplotlib.image.AxesImage object at 0x7f83fc742290>
171 img_vgg16_color_map_show_dimension = visualkeras.layered_view(
172 model,
173 legend=True,
174 show_dimension=True,
175 type_ignore=[visualkeras.SpacingDummyLayer],
176 color_map=color_map,
177 font={
178 "font_size": 61,
179 # 'use_default_font': False,
180 # 'font_path': '/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf'
181 },
182 # to_file="result_images/vgg16_color_map_show_dimension.png",
183 save_fig=True,
184 save_fig_filename="vgg16_color_map_show_dimension.png",
185 )
186 img_vgg16_color_map_show_dimension
<matplotlib.image.AxesImage object at 0x7f83fc7e02d0>
189 img_vgg16_flat_show_dimension = visualkeras.layered_view(
190 model,
191 legend=True,
192 show_dimension=True,
193 type_ignore=[visualkeras.SpacingDummyLayer],
194 draw_volume=False,
195 font={
196 "font_size": 61,
197 # 'use_default_font': False,
198 # 'font_path': '/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf'
199 },
200 # to_file="result_images/vgg16_flat_show_dimension.png",
201 save_fig=True,
202 save_fig_filename="vgg16_flat_show_dimension.png",
203 )
204 img_vgg16_flat_show_dimension
<matplotlib.image.AxesImage object at 0x7f83fc779190>
207 img_vgg16_scaling_show_dimension = visualkeras.layered_view(
208 model,
209 legend=True,
210 show_dimension=True,
211 type_ignore=[visualkeras.SpacingDummyLayer],
212 # min_z = 1,
213 # min_xy = 1,
214 # max_z = 4096,
215 # max_xy = 4096,
216 # scale_z = 0.25,
217 # scale_xy = 5,
218 font={"font_size": 61},
219 # to_file="result_images/vgg16_scaling_show_dimension.png",
220 save_fig=True,
221 save_fig_filename="vgg16_scaling_show_dimension.png",
222 )
223 img_vgg16_scaling_show_dimension
<matplotlib.image.AxesImage object at 0x7f83fc3edd50>
Total running time of the script: (0 minutes 17.625 seconds)
Related examples
Visualkeras: Spam Classification Conv1D Dense Example
Visualkeras: Spam Classification Conv1D Dense Example
