visualkeras: autoencoder example#

An example showing the visualkeras function used by a tf.keras.Model model.

# Authors: The scikit-plots developers
# SPDX-License-Identifier: BSD-3-Clause

# visualkeras Need aggdraw tensorflow
# !pip install scikitplot[core, cpu]

# or
# !pip install aggdraw
# !pip install tensorflow

# pip install protobuf==5.29.4
import tensorflow as tf

# Clear any session to reset the state of TensorFlow/Keras
tf.keras.backend.clear_session()

/home/circleci/.pyenv/versions/3.11.14/lib/python3.11/site-packages/keras/src/export/tf2onnx_lib.py:8: FutureWarning:

In the future `np.object` will be defined as the corresponding NumPy scalar.

encoder Model

encoder_input = tf.keras.Input(shape=(28, 28, 1), name="img")
x = tf.keras.layers.Conv2D(16, 3, activation="relu")(encoder_input)
x = tf.keras.layers.Conv2D(32, 3, activation="relu")(x)
x = tf.keras.layers.MaxPooling2D(3)(x)
x = tf.keras.layers.Conv2D(32, 3, activation="relu")(x)
x = tf.keras.layers.Conv2D(16, 3, activation="relu")(x)
encoder_output = tf.keras.layers.GlobalMaxPooling2D()(x)
encoder = tf.keras.Model(encoder_input, encoder_output, name="encoder")

# autoencoder Model
x = tf.keras.layers.Reshape((4, 4, 1))(encoder_output)
x = tf.keras.layers.Conv2DTranspose(16, 3, activation="relu")(x)
x = tf.keras.layers.Conv2DTranspose(32, 3, activation="relu")(x)
x = tf.keras.layers.UpSampling2D(3)(x)
x = tf.keras.layers.Conv2DTranspose(16, 3, activation="relu")(x)
decoder_output = tf.keras.layers.Conv2DTranspose(1, 3, activation="relu")(x)
autoencoder = tf.keras.Model(encoder_input, decoder_output, name="autoencoder")
autoencoder.summary()

Model: "autoencoder"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ global_max_pooling2d            │ (None, 16)             │             0 │
│ (GlobalMaxPooling2D)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ reshape (Reshape)               │ (None, 4, 4, 1)        │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d_transpose                │ (None, 6, 6, 16)       │           160 │
│ (Conv2DTranspose)               │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d_transpose_1              │ (None, 8, 8, 32)       │         4,640 │
│ (Conv2DTranspose)               │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ up_sampling2d (UpSampling2D)    │ (None, 24, 24, 32)     │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d_transpose_2              │ (None, 26, 26, 16)     │         4,624 │
│ (Conv2DTranspose)               │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ conv2d_transpose_3              │ (None, 28, 28, 1)      │           145 │
│ (Conv2DTranspose)               │                        │               │
└─────────────────────────────────┴────────────────────────┴───────────────┘
 Total params: 28,241 (110.32 KB)
 Trainable params: 28,241 (110.32 KB)
 Non-trainable params: 0 (0.00 B)

Build the model with an explicit input shape

autoencoder.build(
    input_shape=(None, 28, 28, 1)
)  # Batch size of None, shape (28, 28, 1)

# Create a dummy input tensor with a batch size of 1
dummy_input = tf.random.normal([1, 28, 28, 1])  # Batch size of 1, shape (28, 28, 1)
# Run the dummy input through the model to trigger shape calculation
encoder_output = autoencoder(dummy_input)
# Now check the output shape of the encoder
print("Output shape after running model with dummy input:", encoder_output.shape)

# Check each layer's output shape after building the model
for layer in encoder.layers:
    if hasattr(layer, "output_shape"):
        print(f"{layer.name} output shape: {layer.output_shape}")
    if hasattr(layer, "output"):
        print(f"{layer.name} shape: {layer.output.shape}")

Output shape after running model with dummy input: (1, 28, 28, 1)
img shape: (None, 28, 28, 1)
conv2d shape: (None, 26, 26, 16)
conv2d_1 shape: (None, 24, 24, 32)
max_pooling2d shape: (None, 8, 8, 32)
conv2d_2 shape: (None, 6, 6, 32)
conv2d_3 shape: (None, 4, 4, 16)
global_max_pooling2d shape: (None, 16)

from scikitplot import visualkeras

img_encoder = visualkeras.layered_view(
    encoder,
    text_callable="default",
    # to_file="result_images/encoder.png",
    save_fig=True,
    save_fig_filename="encoder.png",
)
img_encoder

2026-01-06 16:56:15.716688: W scikitplot 140172358323072 utils_pil.py:204:load_font] Error loading system font: cannot open resource
2026-01-06 16:56:15.716780: W scikitplot 140172358323072 utils_pil.py:206:load_font] Falling back to PIL default font.
2026-01-06 16:56:15.716891: W scikitplot 140172358323072 _layered.py:216:layered_view] The legend_text_spacing_offset parameter is deprecated andwill be removed in a future release.

<matplotlib.image.AxesImage object at 0x7f7bac2e3190>

img_autoencoder = visualkeras.layered_view(
    autoencoder,
    # to_file="result_images/autoencoder.png",
    save_fig=True,
    save_fig_filename="autoencoder.png",
)
img_autoencoder

2026-01-06 16:56:15.906558: W scikitplot 140172358323072 utils_pil.py:204:load_font] Error loading system font: cannot open resource
2026-01-06 16:56:15.906680: W scikitplot 140172358323072 utils_pil.py:206:load_font] Falling back to PIL default font.
2026-01-06 16:56:15.906808: W scikitplot 140172358323072 _layered.py:216:layered_view] The legend_text_spacing_offset parameter is deprecated andwill be removed in a future release.

<matplotlib.image.AxesImage object at 0x7f7b9408b190>

img_autoencoder_text = visualkeras.layered_view(
    autoencoder,
    min_z=1,
    min_xy=1,
    max_z=4096,
    max_xy=4096,
    scale_z=1,
    scale_xy=1,
    # font={"font_size": 14},
    text_callable="default",
    # to_file="result_images/autoencoder_text.png",
    save_fig=True,
    save_fig_filename="autoencoder_text.png",
    overwrite=False,
    add_timestamp=True,
    verbose=True,
)
img_autoencoder_text

2026-01-06 16:56:16.057579: W scikitplot 140172358323072 utils_pil.py:204:load_font] Error loading system font: cannot open resource
2026-01-06 16:56:16.057785: W scikitplot 140172358323072 utils_pil.py:206:load_font] Falling back to PIL default font.
2026-01-06 16:56:16.057995: W scikitplot 140172358323072 _layered.py:216:layered_view] The legend_text_spacing_offset parameter is deprecated andwill be removed in a future release.
[INFO] Saving path to: /home/circleci/repo/galleries/examples/visualkeras_CNN/result_images/autoencoder_text_20260106_165616Z.png

<matplotlib.image.AxesImage object at 0x7f7b940bb590>

Tags: model-type: classification model-workflow: model building plot-type: visualkeras domain: neural network level: beginner purpose: showcase

Total running time of the script: (0 minutes 3.689 seconds)

Related examples