.. note::
    :class: sphx-glr-download-link-note

    Click :ref:`here <sphx_glr_download_how_to_compile_models_from_keras.py>` to download the full example code
.. rst-class:: sphx-glr-example-title

.. _sphx_glr_how_to_compile_models_from_keras.py:


Compile Keras Models
=====================
**Author**: `Yuwei Hu <https://Huyuwei.github.io/>`_

This article is an introductory tutorial to deploy keras models with Relay.

For us to begin with, keras should be installed.
Tensorflow is also required since it's used as the default backend of keras.

A quick solution is to install via pip

.. code-block:: bash

    pip install -U keras --user
    pip install -U tensorflow --user

or please refer to official site
https://keras.io/#installation


.. code-block:: default

    import tvm
    from tvm import te
    import tvm.relay as relay
    from tvm.contrib.download import download_testdata
    import keras
    import numpy as np







Load pretrained keras model
----------------------------
We load a pretrained resnet-50 classification model provided by keras.


.. code-block:: default


    if tuple(keras.__version__.split(".")) < ("2", "4", "0"):
        weights_url = "".join(
            [
                "https://github.com/fchollet/deep-learning-models/releases/",
                "download/v0.2/resnet50_weights_tf_dim_ordering_tf_kernels.h5",
            ]
        )
        weights_file = "resnet50_keras_old.h5"
    else:
        weights_url = "".join(
            [
                " https://storage.googleapis.com/tensorflow/keras-applications/",
                "resnet/resnet50_weights_tf_dim_ordering_tf_kernels.h5",
            ]
        )
        weights_file = "resnet50_keras_new.h5"


    weights_path = download_testdata(weights_url, weights_file, module="keras")
    keras_resnet50 = keras.applications.resnet50.ResNet50(
        include_top=True, weights=None, input_shape=(224, 224, 3), classes=1000
    )
    keras_resnet50.load_weights(weights_path)







Load a test image
------------------
A single cat dominates the examples!


.. code-block:: default

    from PIL import Image
    from matplotlib import pyplot as plt
    from keras.applications.resnet50 import preprocess_input

    img_url = "https://github.com/dmlc/mxnet.js/blob/main/data/cat.png?raw=true"
    img_path = download_testdata(img_url, "cat.png", module="data")
    img = Image.open(img_path).resize((224, 224))
    plt.imshow(img)
    plt.show()
    # input preprocess
    data = np.array(img)[np.newaxis, :].astype("float32")
    data = preprocess_input(data).transpose([0, 3, 1, 2])
    print("input_1", data.shape)




.. image:: /how_to/compile_models/images/sphx_glr_from_keras_001.png
    :class: sphx-glr-single-img


.. rst-class:: sphx-glr-script-out

 Out:

 .. code-block:: none

    input_1 (1, 3, 224, 224)



Compile the model with Relay
----------------------------
convert the keras model(NHWC layout) to Relay format(NCHW layout).


.. code-block:: default

    shape_dict = {"input_1": data.shape}
    mod, params = relay.frontend.from_keras(keras_resnet50, shape_dict)
    # compile the model
    target = "cuda"
    dev = tvm.cuda(0)

    # TODO(mbs): opt_level=3 causes nn.contrib_conv2d_winograd_weight_transform
    # to end up in the module which fails memory validation on cuda most likely
    # due to a latent bug. Note that the pass context only has an effect within
    # evaluate() and is not captured by create_executor().
    with tvm.transform.PassContext(opt_level=0):
        model = relay.build_module.create_executor("graph", mod, dev, target, params).evaluate()








Execute on TVM
---------------


.. code-block:: default

    dtype = "float32"
    tvm_out = model(tvm.nd.array(data.astype(dtype)))
    top1_tvm = np.argmax(tvm_out.numpy()[0])







Look up synset name
-------------------
Look up prediction top 1 index in 1000 class synset.


.. code-block:: default

    synset_url = "".join(
        [
            "https://gist.githubusercontent.com/zhreshold/",
            "4d0b62f3d01426887599d4f7ede23ee5/raw/",
            "596b27d23537e5a1b5751d2b0481ef172f58b539/",
            "imagenet1000_clsid_to_human.txt",
        ]
    )
    synset_name = "imagenet1000_clsid_to_human.txt"
    synset_path = download_testdata(synset_url, synset_name, module="data")
    with open(synset_path) as f:
        synset = eval(f.read())
    print("Relay top-1 id: {}, class name: {}".format(top1_tvm, synset[top1_tvm]))
    # confirm correctness with keras output
    keras_out = keras_resnet50.predict(data.transpose([0, 2, 3, 1]))
    top1_keras = np.argmax(keras_out)
    print("Keras top-1 id: {}, class name: {}".format(top1_keras, synset[top1_keras]))




.. rst-class:: sphx-glr-script-out

 Out:

 .. code-block:: none

    Relay top-1 id: 285, class name: Egyptian cat
    Keras top-1 id: 285, class name: Egyptian cat




.. _sphx_glr_download_how_to_compile_models_from_keras.py:


.. only :: html

 .. container:: sphx-glr-footer
    :class: sphx-glr-footer-example



  .. container:: sphx-glr-download

     :download:`Download Python source code: from_keras.py <from_keras.py>`



  .. container:: sphx-glr-download

     :download:`Download Jupyter notebook: from_keras.ipynb <from_keras.ipynb>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_